Latest News

A previously unreported Internet-of-Things (IoT) botnet framework, dubbed TuxBot v3 Evolution, has been discovered by cybersecurity researchers. The framework shows signs of being developed with assistance from a large language model (LLM), although the results are not entirely successful.

The LLM was used to generate botnet code, but it included a safety disclaimer that the developer failed to remove before shipping. This suggests that while the AI did aid in constructing the botnet, several functions in the analyzed samples failed to work correctly. A manual code review would have likely resolved these errors, and it’s possible that more polished iterations of the malware exist out there in the wild.

The TuxBot v3 Evolution framework consists of multiple components, including a C-based bot agent that cross-compiles for various architectures (ARM, MIPS, MIPSEL, MIPS64, x86_64, PowerPC, and RISC-V). The Go-based command-and-control (C2) server features a DDoS-for-hire panel, while the custom exploit virtual machine is designed to target vulnerabilities in IoT devices. Additionally, there’s an automated build system and Docker-based test infrastructure.

The bot agent is responsible for brute-force Telnet access on targeted devices using 1,496 credential pairs. It also incorporates exploit code targeting over 30 IoT device families using known vulnerabilities. The C2 server communicates with the bot agent over an encrypted TCP channel, employing a SHA512 domain generation algorithm (DGA) and peer-to-peer gossip protocol with Ed25519-signed commands.

The framework’s modular design allows for flexibility in its operations. It can resort to various fallback mechanisms, including Internet Relay Chat (IRC), DNS TXT queries, and HTTP polling. The lineage of TuxBot v3 Evolution has been traced back to three different botnets: Mirai, AISURU, and Wuhan. Some functions have also been ported from the open-source MHDDoS Python DDoS toolkit.

At least one sample of the malware was uploaded to VirusTotal on January 20, indicating it’s been around for over six months. Evidence suggests that work on the botnet began a year prior, when the author cloned the MHDDoS repository from GitHub. The framework’s description claims it features a professional-grade C2 platform with multi-user admin panel and automated deployment.

The Go-based C2 server component uses three different TCP ports for incoming connections: 1999 (or 31337), 2222, and 9999. These ports handle encrypted command dispatch to connected bots, interactive shell access over SSH, and programmatic interface via JSON, respectively. Once launched, the botnet follows a pre-defined initialization sequence.

This sequence includes loading the C2 address from a multi-tiered architecture with one primary channel and five alternate mechanisms. It also sets up anti-debugging and anti-VM protections to evade analysis tools. The process name is hidden, persistence is installed, and various sub-modules are launched to mount DDoS attacks and establish communication channels over IRC, HTTP, DNS, and P2P.

The dedicated HTTP scanner can manage up to 128 concurrent connections at any given point in time, operating with the goal of discovering vulnerable web interfaces. Persistence is accomplished through systemd service, cron entries, and a watchdog keepalive process to ensure TuxBot remains operational on compromised machines.

Multiple files contain raw LLM chain-of-thought reasoning left verbatim in comments. These comments reveal the internal reasoning as it worked through porting tasks, complete with self-interruptions, decisions, and references to ‘the user’ (meaning the developer who prompted the LLM). This suggests that the AI was used as a judge of sorts, providing guidance on how to develop the botnet.

The core working functions in TuxBot v3 Evolution, coupled with its reliance on AI tools for businesses, signal accelerated integration of features. The framework’s modular design enables what appears to be single developer to come up with a multi-pronged toolset featuring multiple C2 channels and custom exploit VM.

Shared infrastructure with Kaitori v3.9 and AISURU tooling places the TuxBot operator within the Keksec ecosystem, known for running multiple IoT botnet variants in parallel. This variant aims to go beyond Mirai forks by incorporating encrypted C2, DGA, and a modular exploit system.

The disclosure follows recent emergence of two other botnets: RustDuck and AryStinger. These have targeted routers, IP cameras, Android boxes, and poorly secured servers for co-opting them into networks designed to render online services offline and conduct reconnaissance.

Artificial intelligence startup Anthropic is gearing up for a potential initial public offering (IPO) later this year, with bankers lining up investor meetings in advance of the share sale. According to an individual familiar with the plans, Anthropic’s IPO preparations are progressing rapidly as its advisors sound out demand from prospective investors before embarking on a formal roadshow.

The meetings suggest that Anthropic is moving closer to listing on public markets, which could happen as soon as October, although the exact timing remains uncertain. This would mark another significant milestone for the AI industry, following June’s massive SpaceX IPO and further opening up the public markets to companies at the forefront of the AI boom.

Anthropic confidentially filed its IPO prospectus with the Securities and Exchange Commission last month but has yet to disclose when it plans to debut on the stock market. The company was founded in 2021 by a group of executives and researchers who defected from OpenAI over concerns about the direction of their former employer, and has since found early success selling its AI assistants to enterprises.

Anthropic’s popular coding assistant, Claude Code, is one key factor behind its rapid growth, with the company closing a $65 billion funding round at a valuation of $965 billion in May. This puts it above OpenAI’s valuation for the first time and marks another significant milestone for the startup, which has been fueling the AI spending boom.

The involvement of top Wall Street banks Goldman Sachs, Morgan Stanley, and JPMorgan Chase in Anthropic’s IPO planning underscores their growing interest in the AI sector. The three firms are among the biggest on Wall Street by revenue, and their participation highlights the significant role that AI is playing in driving growth for financial institutions as they seek to satisfy investors clamoring for ways to fund or invest in this rapidly expanding theme.

Anthropic’s potential IPO would also mark a shift towards greater transparency and accountability within the industry. The company has been at the center of controversy over export controls on its technology, with the Trump administration recently lifting restrictions on certain products. However, it remains unclear how this will impact Anthropic’s future plans or operations.

The world of model routing isn’t as straightforward as it seems. Our experience in building routing into agentic systems has shown that what appears to be a simple problem quickly becomes a complex challenge.

One dimension that makes model routing surprisingly hard is cost. We expected GPT-4.1 to be cheaper than Claude Sonnet 4.6, but our tests revealed the opposite: Sonnet cost $79 total ($0.19/task) while GPT-4.1 cost $155 ($0.37/task), nearly double.

The explanation for this anomaly lies in caching – a factor that’s often overlooked when discussing routing. Agent workloads tend to reuse large chunks of context across steps, leading to high cache hit rates. When these rates are high, effective input costs drop dramatically.

This experience highlights the importance of considering actual cost, which depends on the interaction between the model, workload, and serving infrastructure. A router that only looks at pricing sheets is optimizing against the wrong numbers – a mistake with significant implications for data analysis tools.

Another dimension that complicates model routing is complexity. A common strategy is to estimate task difficulty and send harder tasks to stronger models, but this approach breaks down in two ways: difficulty can be invisible at routing time, and it’s just one signal among many that needs to be balanced with cost, latency, and other factors.

Secondly, even if task difficulty could be perfectly estimated, it’s not the only consideration. In production environments, routers need to balance multiple competing demands, including compliance requirements, data residency rules, privacy constraints, and approved model lists – all of which can influence where a task should go.

We’ve found that building effective routing systems requires considering more than just cost or quality: latency, reliability, and governance are also crucial factors. This complexity is often overlooked in discussions about routing systems, but it’s essential for data analysis tools that need to handle real-world applications with nuance.

Latency is another challenge when it comes to model routing. It’s tempting to think about latency solely in terms of model size – bigger models are slower, smaller ones are faster – but user experience depends on more than just model speed: infrastructure factors like hardware, cache warmth, and endpoint busy-ness often dominate end-to-end response times.

A theoretically faster model can still produce a slower experience if the serving conditions aren’t right. Routing granularity also plays a crucial role in determining latency: routing once per task adds minimal overhead, but routing at every step introduces additional decision points that increase latency and operational complexity.

These lessons have shaped our approach to building routers for agentic systems – we’ve shifted from treating routing as a classification problem to an optimization challenge. Our algorithm optimizes across cost, quality, and latency simultaneously while staying lightweight enough not to become a bottleneck itself.

The figure below illustrates the result on the AppWorld Test Challenge with a CodeAct agent. Each blue square represents a different configuration of our router, tracing out a cost-accuracy frontier – it shows that the router provides a range of operating points for choosing between cost, latency, or accuracy priorities.

Configuration 1 (latency-optimized) achieves 84% accuracy for $93 and 83s: a 21% cost reduction and 9% latency reduction compared to running Opus alone, with only a 4% accuracy drop. This is what our experience has led us to conclude about routing – it’s not really about choosing models; it’s about optimizing systems.

Models are just one variable – an important but not sole consideration alongside caching behavior, infrastructure state, compliance constraints, and workload patterns. When routing works well, it’s rarely because it found the ‘best’ model for a given task: it’s because it found the best operating point for the entire system.

That’s a harder problem than classification, but it’s the one worth solving – we believe that optimizing systems will lead to more effective data analysis tools. We’ll be sharing more about our technical approach in a follow-up post and look forward to hearing from others who are building routing into their own agentic systems.

Our experience has also shown us that standard difficulty-based routers often fall short, landing in similar accuracy ranges but at higher cost – they don’t explore the full tradeoff space that an optimization-based approach can. Our algorithm’s lightweight nature (roughly 6 ms and 2 kB of memory per task) means it doesn’t become the bottleneck we warned about earlier.

Our router optimizes across multiple factors, including cost, quality, latency, compliance, and reliability – all while staying adaptable to changing workloads and infrastructure conditions. This is a more complex problem than classification, but one that’s essential for building effective routing systems.

We’re sharing our findings in the hope of helping others who are struggling with model routing challenges – if you have any tradeoffs or complexities you’d like to discuss, we’d love to hear about them.

The complexity of model routing is often overlooked, but it’s a crucial consideration when designing data analysis tools. We believe that by prioritizing system optimization over model selection alone, we can build more effective and efficient routing systems – ones that truly meet the needs of real-world applications.

Claude, the popular chatbot from Anthropic, has launched its new reflect feature to help users understand their behavior and optimize how they use the platform. This move comes as concerns about AI usage continue to grow.

The tool is designed to provide detailed insights into how users collaborate with Claude, highlighting key topics, usage patterns, and frequent tasks. Users can review chat activity over periods ranging from one month to a year to gain valuable information – like which features are used most frequently or where interactions tend to slow down.

One of the primary goals of the reflect feature is to encourage users to examine their relationship with Claude more closely. To achieve this, the tool periodically asks questions that prompt users to think critically about their AI use and engage in discussions with Claude.

The insights dashboard offers practical suggestions for using the chatbot effectively without providing repetitive information. The feature also incorporates Anthropic’s proprietary framework for evaluating AI use, which helps determine how to delegate tasks to a chatbot accurately – something experts say is essential when working with AI tools like this one.

Anthropic consulted independent experts in developing the reflect tool, focusing on creating insights that would help users evaluate what works best about Claude for them. The company also drew from youth developmental expertise to inform the design of tools that can assist young adults and parents in understanding their AI use – a key aspect as AI becomes increasingly integrated into daily life.

The reflect feature is currently available in beta form, with both free and paid subscribers who have enabled chat memories able to access it. Users must be at least 18 years old, but Anthropic has taken steps to ensure sensitive conversations are handled carefully – including measures against data breaches and unauthorized access.

Anthropic safeguards user privacy by not incorporating incognito chats into insight reports or accessing files from connected tools like email inboxes or health data. However, reflection insights may cover sensitive topics at a high level, and if Claude previously shared relevant resources during a chat, that information might appear on the dashboard – but only to provide context.

The introduction of this feature marks an effort by Anthropic to give users genuinely useful information about their AI use – something they hope will empower people to make informed decisions. This is particularly important when it comes to using AI tools in business and personal settings alike.

By leveraging data analysis, the company aims to help users interact with Claude more effectively. The goal goes beyond providing a snapshot of past behavior; it encourages users to think about how they can use these insights to improve future outcomes – whether that means streamlining workflows or better assessing AI outputs.

The reflect feature’s development is also an opportunity for Anthropic to collaborate with experts and inform the design of tools that can assist young adults, parents, and businesses in understanding their AI use. As a result, users will be able to access practical insights on how they interact with Claude – making it easier for them to make informed decisions about using these AI tools.

A new state-run artificial intelligence (AI) platform has been launched in California to support government workers. The platform, called Poppy, is a collection of 10 different models tailored to fit the diverse and challenging work of government. This AI assistant was made possible by Executive Order N-12-23 signed by Governor Gavin Newsom late last year.

The development of Poppy involved overcoming several challenges, including privacy, security, governance, and appropriate use considerations. Despite these hurdles, the California Department of Technology (CDT) successfully implemented a consolidated suite of models in just under a year. This achievement is significant given that getting any model up and running within state government can be a daunting task.

According to Shera Mui, deputy director of CDT’s Platform Services, what started as an effort to create a statewide chatbot quickly evolved into something more useful – a platform that could be managed and adapted to new technologies in real-time. ‘We discovered early on that we didn’t want to create large language models; that was not going to be the smart thing to do,’ Mui explained.

The team at CDT took a different approach, focusing on creating a platform that would allow users to securely choose a model best suited for their task. This flexibility is particularly useful in government work, where tasks can range from deep analysis of legal documents to creating code for new websites. The platform’s ability to keep all user data and information secure has been a major driving factor behind its development.

The early access phase involved giving 70 departments with around 100 users each the opportunity to test the first iteration of Poppy. This pilot program helped drive discussion around potential use cases, which in turn informed the development of subsequent models. The platform’s flexibility is one of its key strengths – it allows users to choose from multiple models without having to duplicate data or compromise on security.

Dr. Lucy Andrews, a scientist with the California Department of Water Resources (DWR), has seen firsthand the impact that Poppy can have in labor- and time-intensive work. She noted that the tools have played a critical part in determining the intellectual and financial value of her department’s contributions across various sources. This is particularly important for DWR, which invests tens of millions of dollars annually in scientific research.

Previously, tracking these contributions was a manual process that involved weeks of data entry using an Excel spreadsheet. With access to AI tools like Poppy, this time-consuming task has been significantly reduced – users can now complete it within 1-2 hours with a slow internet connection. The ability to track and analyze scientific relationships more effectively is also a major benefit for DWR.

Andrews noted that the department’s focus on accountability means they are taking a conscientious approach to model selection and use cases before implementing Poppy. This includes considering the environmental impact of running data centers, which support AI operations. While there may be some concerns about energy consumption, staff is working to minimize this by carefully selecting models and optimizing their usage.

Early metrics from the statewide launch show that users are interested in document summaries, policy analysis, and complex code creation – all key areas where Poppy’s capabilities can make a significant difference for government workers. Some departments have also expressed interest in exploring image generation features as part of future additions to the platform. However, more work needs to be done on outlining a long-term roadmap for these developments.

As for what’s next for Poppy, Mui sees opportunities for further expansion and improvement. This includes adding new features, tailoring models to specific departmental needs, and streamlining data sets to enhance cooperation among departments. CDT has requested $1 million in funding during the budget process toward this end – a significant investment that reflects the potential of AI tools like Poppy.

It’s worth noting that while Poppy is an advanced toolset for government workers, no decision is left solely to an AI system. Humans maintain control over all outcomes and are responsible for ensuring that AI-generated information aligns with their goals and objectives.

HitPaw Edimakor is marking its second anniversary with a major promotion, offering creators around the globe up to 60% off on VIP plans and free gifts with selected purchases. The limited-time sale provides an opportunity for content makers to access Edimakor’s powerful AI tools at exceptional prices.

The platform has gained popularity among creators due to its comprehensive suite of features, which simplify every stage of content creation. Unlike traditional video editors, Edimakor is an all-in-one AI creative platform that enables users to generate, edit, and export professional content in one place.

Edimakor’s AI capabilities are impressive, with several advanced tools at their disposal. These include the AI Image Generator powered by Seedream 5.0 Pro, NanoBanana Pro, and GPT-Image 2 for creating high-quality images from simple prompts. The platform also features an AI Video Generator integrated with various AI models for producing cinematic videos.

Additionally, Edimakor offers AI Text-to-Speech with more than 1,000 natural voices across 80+ languages, as well as an AI Subtitle Generator that automatically creates and translates subtitles into over 130 languages. The platform also includes an AI Music Generator that turns text, poems, or creative ideas into original music.

The Easy Video Editor is another key feature of Edimakor, providing drag-and-drop editing tools that make professional video creation simple for beginners and experienced creators alike. This versatility makes it an attractive option for businesses looking to leverage AI tools in their content creation process.

Edimakor’s anniversary promotion includes several exciting deals for creators. The platform is offering 60% off on VIP plans, with a perpetual plan available for $55.95, including 10,000 lifetime AI credits plus 2,000 bonus credits. A one-year plan can be purchased for $39.99, including 3,000 annual AI credits plus 2,000 bonus credits.

AI Credits are also discounted during the anniversary sale, with up to 65% off on packages of varying sizes. For instance, a package of 100,000 credits is available for $203.99 (1-year subscription), while 10,000 credits can be purchased for $23.99 (1-month subscription).

Another exclusive offer includes the Edimakor Screen Recorder, which will be given away free to customers who purchase any Edimakor Video Editor Perpetual plan during the anniversary event.

The limited-time offers are a great opportunity for creators to access professional AI tools at outstanding prices. With its comprehensive suite of features and user-friendly interface, Edimakor is an attractive option for businesses looking to streamline their content creation process using AI technology.

Edimakor’s 2nd anniversary sale won’t last forever, so it’s essential for creators to visit the platform today and claim their exclusive savings. By doing so, they can start creating their next project with the latest AI tools at a fraction of the cost.

CNBC’s Jim Cramer has weighed in on the current state of the stock market, dismissing concerns that it is experiencing a bubble similar to the one preceding the dot-com crash. According to Cramer, companies like SpaceX are exceptions rather than representative of the broader market.

While some investors have raised eyebrows at the rapid gains made by semiconductor and AI-related companies, Cramer argues that these outliers do not reflect the overall market conditions. He points out that there is ‘some froth’ in certain areas but emphasizes that this does not accurately represent what drives the market’s performance.

The past year has seen significant stock price increases as enthusiasm for artificial intelligence continues to grow. Companies like Micron and Sandisk have experienced massive gains, with their shares rising by over 243% and 644%, respectively. This surge in prices has led some investors to question whether the market is becoming overheated, drawing comparisons to the dot-com boom of the late 1990s.

Cramer disagrees with this assessment, citing lower interest rates as a key factor contributing to the current market conditions. He also notes that corporate earnings are stronger than they were during the tech bubble and points out that valuations are more reasonable compared to those seen in the past.

The latest consumer price index report came in below expectations on Tuesday, which has eased concerns about potential rate hikes by the Federal Reserve. Cramer believes this development reduces the likelihood of a series of significant interest rate increases, similar to what occurred before the dot-com crash.

New Fed Chair Kevin Warsh’s comments on Tuesday also provided reassurance for investors. According to Cramer, Warsh did not indicate that he would tighten monetary policy if inflation remains at current levels. This suggests that the market is unlikely to experience a sharp correction in the near future.

Cramer also draws attention to valuations of major companies trading at what he considers attractive multiples despite reporting strong results. Bank of America, Goldman Sachs, and JPMorgan all reported substantial earnings and revenue beats on Tuesday, with their shares trading at roughly 12-18 times forward earnings. Cramer’s Charitable Trust owns shares in these companies.

These valuations are significantly lower than those seen during the peak of the dot-com era, when the S&P 500 traded at over 25 times forward earnings. According to FactSet data, this multiple is now around 20, which Cramer considers more reasonable but not cheap by any means.

Cramer also points out that several technology companies are trading at relatively low multiples compared to their historical norms. SK Hynix and Micron have shares priced at roughly four and six times forward earnings estimates for 2027, respectively. Nvidia’s multiple is similar to the broader market despite its dominant position in artificial intelligence.

What characterizes this market, according to Cramer, is the relatively low valuation of many large-cap stocks.

Pattern recognition is one of the most powerful tools humans possess. It’s a fundamental aspect of our cognitive structure, allowing us to quickly respond to threats and identify patterns in data. However, this ability also makes us prone to errors, as seen in pareidolia – the phenomenon of seeing patterns that aren’t really there. We’ve all been guilty of it at some point: seeing faces in rocks or finding meaning in song lyrics when none exists.

Artificial intelligence (AI) relies heavily on pattern recognition, using machine learning methods to power through vast amounts of data and identify significant patterns. But new research suggests that AI’s pattern recognition abilities are not foolproof – they can be easily fooled by out-of-distribution samples. This is particularly concerning when it comes to detecting life beyond Earth.

The study, titled ‘Can AI Detect Life? Lessons from Artificial Life,’ was conducted by Ankit Gupta and Christoph Adami from Michigan State University. They used the Avida Digital Evolution Platform (Avida) – an artificial life software platform that allows researchers to create digital organisms and study evolutionary biology. The researchers generated tens of thousands of digital organisms, some containing instructions for self-replication and others not.

Their goal was to train a neural network to recognize whether these digital organisms were living or non-living based on their molecular structure. They achieved an impressive 99.7% accuracy with the initial sample set. However, when they introduced out-of-distribution samples – molecules that weren’t part of the original dataset – the AI became confused and began misclassifying them as living.

The researchers found that it took as few as 150 tweaks to the code for the AI to confidently proclaim that a non-living organism was indeed alive. This is concerning, especially when considering future missions to Mars or other planets where AI will be used to detect signs of life. The likelihood of encountering out-of-distribution samples in these environments is substantial.

Christoph Adami, co-author and professor at Michigan State University’s departments of microbiology and molecular genetics, as well as physics and astronomy, emphasized the importance of human oversight when using AI for critical tasks like detecting life. ‘You need an independent way of checking their work,’ he said. ‘There needs to be a human in the loop.’ This is particularly challenging on space missions where communication with Earth can be delayed or unreliable.

The researchers’ next step will be to train their AI on real-world data and test its ability to detect life beyond what it’s been trained on. However, this study highlights a crucial vulnerability in current AI methods – their susceptibility to out-of-distribution high-confidence failures. This could have significant implications for future astrobiology missions.

The use of AI-generated images or LLMs as judges has become increasingly common in various fields, including data analysis and machine learning jobs. While these tools can be incredibly powerful, they are not infallible. As seen in this study, even the most advanced AI systems can be easily fooled by out-of-distribution samples.

The researchers conclude that if false positives outnumber true positives in extraterrestrial measurements due to AI’s propensity for being misled by out-of-distribution samples, we risk accepting high-confidence classifications at face value. This emphasizes the need for a fact-checker or human oversight when using AI for critical tasks like detecting life beyond Earth.

The study’s findings have significant implications for future astrobiology missions and highlight the importance of developing more robust data analysis tools that can handle out-of-distribution samples. As we continue to rely on AI in various fields, it’s essential to acknowledge its limitations and develop strategies to mitigate these vulnerabilities.

A proposal has been put forward by Google DeepMind’s chief, Demis Hassabis, calling for a US-led body to test frontier artificial intelligence models. The proposed organization would cover both domestic and international developments in the field, including open and closed systems.

The aim of this initiative is to identify potential dangers and cybersecurity risks associated with these advanced AI models. It also seeks to determine whether existing safeguards can be bypassed, which could have significant implications for their deployment.

Hassabis has warned that artificial general intelligence capable of matching human brain function across a range of tasks may only be a few years away. This raises concerns about the need for oversight and regulation in this area, with society having a ‘precious window’ to establish effective controls.

The proposed framework would draw inspiration from existing regulatory bodies such as FINRA, which oversees US securities firms. Hassabis believes that a US-initiated system could become the foundation for international standards on AI testing and deployment.

A key advantage of this approach is that it would provide a common testing process for AI companies, avoiding the need for separate licensing regimes for each model. Independent testers would be able to compare models against established thresholds and update tests as new capabilities emerge.

The White House has already taken steps towards pre-release oversight, with an executive order issued in June seeking voluntary access to frontier models before their release. This move was followed by agreements from Google DeepMind, Microsoft, and xAI to provide models for federal national security testing.

However, the debate around AI regulation continues, with concerns raised over the recent restriction on access to Anthropic’s Mythos models due to safety control bypass risks. While Anthropic complied with this decision, they argued that it did not justify a broad recall of their models.

The US Congress is also considering additional requirements for frontier developers, including reporting dangerous capabilities and breaches within seven days of discovery. This move marks a shift from debating the need for oversight to deciding who conducts tests, which models qualify, and whether regulators can delay deployment.

A new development in the world of artificial intelligence (AI) is set to revolutionize how organizations manage their workforce. Betterworks, a leading provider of performance management solutions, has launched its Model Context Protocol (MCP) Server, enabling secure access to performance data for AI assistants such as ChatGPT and Microsoft Copilot.

The MCP capabilities allow these AI agents to securely access Betterworks’ performance data, helping organizations transition from reactive work to proactive execution. This marks a significant shift in the way businesses use AI tools for business, moving beyond mere response to prompts and towards more proactive decision-making.

According to Doug Dennerline, CEO of Betterworks, ‘the next generation of AI will understand what’s happening across your business and proactively surface the insights that need your attention.’ By providing trusted performance intelligence, leaders can spend less time searching for information and more time coaching their teams, making better decisions, and driving business results.

Many organizations already use AI assistants to prepare for performance reviews, summarize information, and identify workforce trends. However, until now, Betterworks data had to be manually gathered before these tools could deliver meaningful insights. This new development streamlines the process, allowing leaders, managers, and HR teams to simply ask natural-language questions about goals, recognition, team performance, and organizational insights directly from their AI assistant.

For example, a manager can ask: ‘Help me prepare for my 1:1 with Jordan, including goal progress…’ or ‘Which of my team goals are at risk this quarter?’ Instead of spending time gathering information, Betterworks helps managers prepare for coaching conversations, identify emerging risks, and generate leadership-ready updates. By making trusted performance context instantly accessible, organizations can move faster, respond sooner, and make more confident talent decisions.

Security remains a top priority with the MCP Server. Every interaction respects Betterworks’ existing authentication and native permission model, ensuring users only access the information they already have permission to see. Organizations can confidently extend sensitive performance data into AI workflows without introducing new access controls or compromising governance.

The initial release provides secure access to Betterworks goals, teams, users, recognition, and hashtags, with additional capabilities and use cases to be supported as releases follow throughout the year. As organizations increasingly adopt AI across the workplace, Betterworks is making trusted, current performance and talent intelligence available wherever decisions happen—helping leaders act with greater speed, confidence, and context.

The MCP Server is currently available in beta for customers, marking an important step towards integrating data analysis tools into everyday business operations.

Senate Banking Committee Chair Tim Scott has expressed interest in hearing Federal Reserve Chairman Kevin Warsh discuss data centers and artificial intelligence during a scheduled appearance before the committee this week.

The conversation is set to take place as part of Warsh’s semiannual monetary policy report to Congress, but Scott hinted that he wants to explore broader topics beyond traditional Fed matters. When asked what he hopes to hear from Warsh on CNBC’s Squawk Box, Scott emphasized the importance of addressing artificial intelligence and data centers in South Carolina.

Specifically, Scott is concerned about the impact of data centers on local electricity and water usage. He noted that some parties in South Carolina are pushing for a ban on these facilities due to concerns over increased utility bills, which has sparked a national trend with various states implementing moratoriums on new data center development.

The issue extends beyond state borders, as Scott framed it within the context of global competition between the US and China. He believes that artificial intelligence will play a crucial role in determining which country emerges victorious, and he wants to ensure that the US is positioned for success by addressing its own challenges at home.

Scott’s comments suggest that he sees data centers and AI as critical components of America’s future competitiveness, particularly when it comes to innovation and economic growth. By engaging with Warsh on these topics, Scott aims to gain a deeper understanding of how to balance the benefits of technological advancement with concerns over resource usage and local impact.

Hachette Book Group, Cengage Learning, and Elsevier have filed a federal class action suit against Google over allegations that the tech giant has been using copyrighted literature to train its artificial intelligence platform Gemini. The publishers claim that Google scraped their works from online pirating sites and licensed databases without permission, then used them to generate AI-created copies of books and articles.

The lawsuit alleges that Google’s actions will ‘weaken the incentive to create’ by allowing the tech giant to profit from copyrighted materials without compensating authors or publishers. The suit charges Google with four counts of copyright infringement and claims that the company has assembled a vast trove of copyrighted materials through its Gemini AI system.

Google obtained the copyrighted material by scraping online pirating sites and licensed databases, including those used for its Google Books and Google Scholar platforms. However, it was specifically forbidden from copying these works for other purposes, yet allegedly did so many times over to train its multi-billion-dollar generative AI system.

The publishers claim that Google deployed a purpose-built service designed to generate content that creates direct substitutes for their original work. This includes allowing users to type requests into Gemini and receive full or large portions of text from novels and articles, as well as generating new books.

AI-generated works can then be sold, competing directly with the publishers’ material. The suit claims that these substitute works take multiple forms, including verbatim copies of entire works, replacement chapters of academic textbooks, summaries, alternative versions, and inferior knockoffs that copy creative elements of original works.

Gemini’s outputs are tailored to mimic the expressive elements and creative choices of specific authors, making it difficult for publishers or authors to compete with. For example, Gemini can generate a 100-page murder mystery in just 20 minutes for $0.39 – an unprecedented scale and speed that displaces legitimate sales of books and journal articles.

Internal documents cited in the suit show that Google employees noted the potential risks of using copyrighted materials without permission or compensation. Despite this, Google allegedly did so anyway, flagging internally that it was ‘highly problematic’ but proceeding with the use of these works to train its AI models.

The publishers claim that Google could have simply bought the rights to copy the work and paid them accordingly, making this a classic case of copyright infringement. While AI technology may be new, the legal principles at the center of this case are not – copyright law applies equally to all companies, including those using novel technologies like Gemini.

This lawsuit joins a growing number of publishers and authors suing artificial intelligence companies over similar copyright infringement claims. Hachette has also filed joint suits with McGraw-Hill and MacMillan Publishing against Meta’s AI in May, highlighting the increasing concern about AI-generated content competing directly with human-created works.

Demis Hassabis, the chief of Google’s artificial intelligence division and Nobel laureate, has called for the United States to spearhead a standards body that would oversee new AI models and assess national security risks. In an article posted on X, Hassabis emphasized the need for ‘urgent action’ to address the challenges associated with artificial general intelligence (AGI), which refers to the point at which AI matches or surpasses human intelligence.

Hassabis noted that frontier models have already posed significant cybersecurity threats, and other potential dangers such as nuclear and biological risks may soon emerge as capabilities continue to advance. He proposed a US-led public-private partnership overseen by the federal government as a solution to help tackle these threats. The White House, State Department, and Department of Commerce have been approached for comment.

The comments come on the heels of recent calls among industry leaders for an AI watchdog. Despite growing concerns over regulation, leading AI models are increasingly being subject to restrictions from public and private sectors alike. For instance, Anthropic was locked in negotiations with officials after the Trump administration temporarily imposed export controls over an advanced model, while OpenAI faced similar restrictions as it was initially requested by the US government to limit the rollout of a new model.

Hassabis argued that the US is well-positioned to lead in developing an AI framework ‘given its economic and technical standing.’ He suggested establishing a new Standards Body modeled on a federally overseen public-private partnership or self-regulatory organization, similar to the Financial Industry Regulatory Authority (FINRA), which regulates brokerage firms and exchange markets in the US. The proposed body would need substantial funding to attract world-class technical talent and provide necessary compute resources for large-scale testing.

Funding would likely come from industry, Hassabis said. Frontier labs would initially voluntarily share models with the Standards Body for review up to 30 days before release; after that, sharing would become mandatory for deployment in the US market if shown to be effective. The proposed body could also leverage specific agentic AI tests to identify attempts to bypass safety guardrails or signs of deception and ensure best practices such as digitally watermarking AI-generated images.

The calls for greater regulatory oversight come amid a heated competition between the US and China to develop and deploy AI models. Recent model releases from Chinese companies, including DeepSeek and Z.ai, are seen by many as highly competitive compared to leading frontier systems from Anthropic and OpenAI. As a result, US lawmakers are currently considering how to curb the growing adoption of Chinese AI models by homegrown companies, which raises ‘serious concerns’ according to the State Department.

Hassabis’s proposal for an AI standards body has been echoed by other industry leaders, including Anthropic CEO Dario Amodei and OpenAI’s Sam Altman. The two called for a US-led coalition to shape rules and standards around AI at a G7 meeting with tech leaders and heads of state that included President Donald Trump earlier this month.

The need for an AI watchdog has become increasingly pressing as the development and deployment of advanced models continue to accelerate. Hassabis emphasized the importance of addressing national security risks associated with AGI, which he believes requires ‘urgent action.’ The proposed Standards Body would provide a framework for overseeing new AI models and assessing potential dangers, ensuring that best practices are followed in developing and deploying these technologies.

A proposal from Google DeepMind’s Demis Hassabis has sparked a significant conversation about the need for stricter regulation in the field of artificial intelligence. The plan, which would see a new regulatory body established to oversee the development and deployment of advanced AI models, carries weight across various sectors, including tech investing and cryptocurrency.

Hassabis, co-founder and CEO of Google DeepMind, has been advocating for greater oversight of AI development. In an interview with Axios, he outlined his vision for a regulator that would be funded by the industry itself, staffed by top technical talent, and ultimately accountable to Washington. This watchdog wouldn’t simply issue reports; it would have the authority to screen advanced models before release and coordinate an industry-wide slowdown if necessary.

The proposal is part of a broader effort to address concerns about AI safety. Hassabis has warned that ‘race conditions’ in AI development – where competitive pressure pushes labs to ship faster than they can verify safety – could lead to catastrophic consequences. He believes that pre-release testing requirements are essential, and companies should be required to prove their models are safe before releasing them to the public.

Hassabis’s proposal is not a sudden departure from his previous views on AI regulation. In 2026, he publicly expressed concerns about the risks associated with rapid AI development. His warning was prompted by the prospect of Artificial General Intelligence (AGI) emerging as early as 2029 – a milestone that would see AI match or exceed human-level reasoning across domains.

Hassabis’s call for greater regulation is significant because it comes from within the industry itself. As CEO of one of the world’s leading AI labs, he is essentially asking for constraints on his own company’s operations. This lends credibility to his proposal and underscores the need for a more systematic approach to AI regulation.

However, some have raised concerns that Hassabis’s plan could favor established players with deep pockets over smaller competitors. The proposed regulatory regime would require expensive pre-release testing and employ world-class technical reviewers – costs that startups and open-source developers might struggle to bear. This raises questions about the potential impact on innovation in the field.

Hassabis has emphasized the importance of global cooperation on AI safety, but his proposal is anchored in American governance. By doing so, he’s making a geopolitical bet: that the US is uniquely positioned to lead this effort due to its institutional credibility, technical talent pool, and market influence.

Siri, Apple’s virtual assistant, is set for a major overhaul as the company seeks to dominate the market for personal assistants. The revamped AI will leverage its massive user base and deep integration with personal data stored on devices to compete directly with ChatGPT and other top-tier AI assistants.

The key to Siri’s success lies in its ability to tap into users’ existing ecosystems, providing a seamless experience that rivals those of competitors. According to analysts at eMarketer, Apple’s scale is unmatched, with billions of active devices already in users’ hands. This extensive reach allows the company to infuse personalized AI features designed around its users.

Apple’s approach differs from the general free-for-all model employed by other companies. Instead, it focuses on creating a safe and highly personalized experience that leverages users’ trust. As analyst Gadjo Sevilla noted in a recent episode of ‘Behind the Numbers,’ Apple is not competing directly with ChatGPT but rather offering its own brand of AI features.

The upgraded Siri will be able to handle complex requests like adding photos to emails or showing specific images from user accounts. This level of functionality was demonstrated at Apple’s developer conference last month, where the company showcased the assistant’s capabilities in action. The revamped AI is expected to launch two years after initial promises and will incorporate Google’s Gemini AI model and cloud technology.

Apple’s control over both hardware and software provides opportunities for AI integration that third-party developers can’t replicate. However, this advantage initially extends only to Apple’s own apps and services, potentially limiting functionality with third-party applications like Gmail or WhatsApp. Analyst Grace Harmon noted that users who rely heavily on non-Apple services may find Siri’s capabilities limited.

Scale and trust are the two fundamental strengths positioning Apple favorably in the AI race. As eMarketer analyst Grace Harmon emphasized, ‘The two big things that are really working in Apple’s favor are scale and trust.’ With billions of active devices at its disposal, Apple has an almost unimaginable distribution network for AI features.

Apple is investing heavily in research and development (R&D) to support the revamped Siri. The company is using foundational models developed in combination with Gemini, as well as its own on-device processing capabilities. Analyst Gadjo Sevilla noted that Apple isn’t trying to reinvent the wheel but rather build upon existing AI technology.

The key differentiator for Siri AI lies in its ability to access and process personal information directly on users’ devices. This creates context awareness that competitors struggle to match, allowing Siri to provide a more personalized experience. Analyst Gadjo Sevilla explained that ‘A lot of what Siri AI is going to be able to do for its users securely will be on device.’ The assistant can pull answers from old message threads and extract information from emails without opening any apps.

Apple’s pitch centers around privacy, emphasizing the need for Siri to access messages, mail, and photos. The company claims it handles what it can directly on devices, with anything sent to the cloud used only to answer requests before being deleted. Analyst Grace Harmon noted that personalization is a crucial aspect of AI assistants, allowing users to tap into their existing data.

Despite its advantages, Apple faces several obstacles in the AI assistant race. The company has historically been criticized for delayed investment and limited third-party integration. As analyst Grace Harmon pointed out, ‘For a long time Apple has been kind of over promising, under delivering.’ This lack of urgency may hinder Apple’s ability to keep pace with competitors like Google or Microsoft.

Hardware limitations mean Siri AI won’t be available on all devices initially, potentially creating a fragmented user experience. Early demonstrations also suggest the assistant takes time to process requests, indicating it’s still in development ahead of its September launch. Third-party developers may face challenges accessing Siri AI integrations, which could further limit functionality.

The revamped Siri is set to put fully capable AI within reach of hundreds of millions of people who have never opened ChatGPT. However, users relying heavily on non-Apple services like Gmail or Google Photos may find Siri’s capabilities limited until those apps allow deeper integration.

The summer vacation season is a time for tech companies to think creatively about their workforces. With employees taking time off, it’s essential to keep the workflow running smoothly. For years, this has meant transferring urgent tasks between colleagues or creating overloads for those who remain in the office.

But 2026 marks a shift in approach as technology companies increasingly turn to advanced AI agents to support human teams and ensure business continuity. This change is significant, with many tech giants investing heavily in these digital assistants.

At WSC Sports, AI assistants are seen as an evolution of the digital workplace. The company’s internal pilot aims to evaluate the integration of AI agents into their organizational workflow. Shai Diament, VP of account management at WSC Sports, emphasizes that the ultimate goal is to ease employee burdens and improve professional quality of life.

The concept of AI assistants has transformed from its passive roots. Today’s advanced agents are proactive team members that listen, observe, and integrate seamlessly into organizational workflows. They can surface solutions, insights, and tasks independently in real-time without requiring human intervention.

‘One of the greatest contributions of this technology is allowing employees to truly disconnect during their vacations,’ Diament explains. By doing so, it serves as a vital support anchor for teams, maintaining a steady workflow even when employees are away on summer vacation or peak periods.’

The integration of AI agents is being tested in various roles and departments within WSC Sports’ internal environment. The company aims to closely monitor system performance and identify which positions and interactions deliver the highest value, responsiveness, and support to human teams managing it.

‘Our vision is to execute a broader roll-out,’ Diament says. ‘We want to expand these agents into additional roles across the organization and embed them in our daily operational infrastructure.’ This would alleviate employee workload and drive company-wide efficiency.

At SeatPick, another tech company, AI assistants are taking on a more tangible form with their agent named Naftali. Developed by global ticket resale price comparison platform SeatPick, Naftali assists human teams with both software development tasks and information gathering and analysis.

‘One of Naftali’s biggest advantages is helping us maintain business continuity during non-standard hours,’ Guy Kogel, CTO and co-founder at SeatPick, explains. This includes periods when the company experiences massive traffic due to events like World Cup matches or holidays.’

Naftali serves as a force multiplier by allowing existing workforces to remain productive and stress-free, especially during off-hours or peak periods. Kogel emphasizes that ‘he helps employees perform research and investigations much faster,’ enabling engineering teams to focus on higher-impact projects while other teams can move forward with improvements.

The integration of specialized AI agents is also critical in highly regulated fields like healthcare technology. At Navina, an intelligent agent named Ofir supports service teams for primary care physicians in the US. When a complex question arises regarding a medical recommendation made by the system, Ofir dives deep into full workflow documentation and analyzes calculation steps.

‘AI agents help bridge availability gaps during peak seasons or vacation periods,’ Shlomit Labin, VP AI at Navina explains. They shorten response times, reduce bottlenecks, and allow human teams to focus on truly complex cases while maintaining a continuous workflow and high quality of service.’

At Plarium, the focus is on creating a shared memory layer that acts as a bridge for the entire team. This ensures workflows build upon past achievements rather than repeating them. According to Tomer Daniel, AI product leader at Plarium, ‘it’s essential not to erase employees’ memories every night,’ just like you wouldn’t hire someone brilliant and then forget their work.

Daniel explains that most AI tools behave like this: every conversation starts from scratch with no idea what the team already knows. At Plarium, they are building Second Brain - a shared memory layer that turns documents, decisions, and hard-won lessons into knowledge its AI assistants can use.

‘When a company remembers, it stops reconstructing the past,’ Daniel notes. ‘Get to build on it.’ The potential impact of this technology could be significant for companies looking to streamline operations and improve employee productivity - we’ve seen early results from Plarium’s internal testing with Second Brain.

According to Labin at Navina, AI agents are crucial in fields like healthcare where maintaining business continuity is critical. With the help of AI tools, medical teams can focus on complex cases while ensuring a continuous workflow.

A federal judge in California just threw out a proposed class-action lawsuit accusing Google of secretly monitoring users’ private emails, chats, and video calls through its Gemini AI assistant. The ruling comes after the plaintiffs failed to provide enough detail about what information was accessed or how they were harmed by the alleged tracking.

The suit, filed by Thomas Thele and Melo Porter in November, accused Google of violating state and federal wiretapping statutes and invading users’ privacy when it switched on Gemini as a default feature for all Gmail, Chat, and Meet users in October 2025. Before that, users had to actively opt-in before Google’s AI could interact with their accounts.

Google argued that the plaintiffs never adequately alleged that their own data was touched or described a concrete enough injury to satisfy the standing requirements under Article III of the Constitution. The company pointed out that the plaintiffs failed to explain when they first signed up for Google’s services, whether they were already using Gmail, Chat, or Meet before Gemini was enabled by default on October 10, 2025.

The judge agreed with Google’s motion to dismiss the amended complaint, concluding that it failed to describe with sufficient detail the central harm it was built around: Google’s supposed intrusion into users’ private communications. The opinion highlighted several specific gaps in the complaint, including a lack of explanation about why October 10, 2025, was significant.

The judge noted that the plaintiffs never explained whether they were shown Google’s privacy policies when creating their accounts or confirmed whether Gemini’s features were already active at account setup. There was also no statement on whether users had since turned off the AI assistant’s features. The opinion stated that nothing in the complaint ruled out the possibility that Gemini had been running as a default feature well before October 2025.

Judge Wise found that the plaintiffs failed to show their own personal data was actually affected by Gemini, despite broadly describing sensitive financial, medical, and employment information that can live inside a Google account. The complaint didn’t point to any specific messages or communications that Gemini supposedly analyzed nor identify any particular personal data used by the AI tool.

The judge rejected the plaintiffs’ bid to pursue an order blocking Google’s practices going forward, finding they hadn’t shown they or other members of the proposed class faced an ongoing or future risk of the same harm. The court noted that users can eliminate any risk by switching off Gemini’s ‘smart’ features in their account settings.

Despite dismissing the complaint, Judge Wise gave the plaintiffs 21 days to file a new version addressing the shortcomings. Citing the general principle that courts should freely allow amended pleadings when doing so serves the interests of justice and helps resolve cases on their merits rather than procedural technicalities, the judge allowed the plaintiffs another chance.

The underlying complaint accused Google of violating California’s constitutional privacy protections and the state’s Invasion of Privacy Act, which bars secretly recording or intercepting confidential communications without consent. It also alleged violations of California’s computer data access law and the federal Stored Communications Act, both of which prohibit intentionally accessing protected electronic information without authorization.

Google argued that users can opt-out of Gemini’s features by switching them off in their account settings. The company pointed out that this option is available to all users who want to avoid having their private communications monitored or analyzed by the AI assistant. Users have control over how they use the Gemini tool and whether it interacts with their accounts.

The ruling comes as a significant development in the ongoing debate about artificial intelligence assistants and data analysis tools used for businesses. Concerns are being raised about user privacy and consent as more companies integrate AI into their services.

The Supreme Court recently ruled that police conducted a Fourth Amendment search when they obtained a robbery suspect’s Google Location History data through a geofence warrant. This decision strengthens privacy protections for cell-phone location records while leaving key questions about the warrant itself unresolved.

Amazon is facing a class-action lawsuit over claims its Ring doorbell cameras used facial recognition technology to scan, identify, and store people’s faces without consent. Similar concerns are being raised about user privacy in this case.

A group of Black employees who accused Google of maintaining disparities in hiring, pay, and promotions has reached a settlement with the company. The lawsuit alleged that Google engaged in a ‘pattern and practice’ of discrimination against its minority workers.

This case highlights the need for companies to be transparent about their use of AI tools and data analysis techniques. Companies like Google must consider the potential consequences of using these technologies without proper safeguards in place.

A Florida father has filed a federal lawsuit against Google alleging that its Gemini AI assistant contributed to events leading up to his son’s suicide and an alleged attempt to stage a violent incident near Miami International Airport. The lawsuit raises concerns about the potential consequences of using AI assistants without proper safeguards in place.

At the forefront of innovation in education, a significant development has taken place with the launch of ATL Saathi. This is a web application powered by Gemini, designed to provide educators at Atal Tinkering Labs (ATL) with a 24/7 planning and training assistant. The tool aims to transform these labs into AI-Augmented Discovery Labs, marking a crucial step towards leveraging technology in education.

The initiative builds upon the existing work of ATL, which has been providing access to new technologies such as 3D printing, IoT, and robotics for over 1.1 crore students across India. However, with this new development, the focus shifts from mere access to physical lab infrastructure to driving meaningful outcomes like accelerated innovation and enhanced learning metrics.

The collaboration between Atal Innovation Mission (AIM) and Google DeepMind has been instrumental in bringing about this change. During the AI Impact Summit in February 2026, it was announced that Google would help incorporate robotics and coding into local curricula, integrate Gemini thoughtfully into teacher workflows, and build a safely guardrailed AI assistant for students grounded in national curriculum standards.

According to experts, behind every good student is a great teacher. This understanding has guided the efforts of both AIM and Google in developing ATL Saathi. The tool’s primary goal is to support educators by streamlining their workload and providing them with the necessary tools and digital skills required for today’s classrooms.

The development process involved close collaboration between AIM, NITI Aayog, and Google. This ensured that the tool was grounded in teacher needs, accurately reflected ATL’s educational principles and pedagogy, and created genuine value for ATL educators. The result is a comprehensive platform designed to empower teachers and enhance student learning outcomes.

ATL Saathi boasts several key features that make it an invaluable resource for educators. Firstly, it provides streamlined onboarding and content curation through NotebookLM. This ensures that educators always have access to the most up-to-date content, including summarized modules, AI-generated infographics, video overviews, and interactive quizzes for 12 core modules from the ATL prescribed curriculum.

This micro-learning approach replaces lengthy videos, allowing teachers to quickly familiarize themselves with complex topics. Additionally, the tool offers advanced project generation capabilities through its ‘Push & Pull’ mentorship feature. For 10 core modules, teachers can access an interface that supports both generating project ideas and providing detailed experiments for students who bring their own problem statements.

Multilingual accessibility is another significant aspect of ATL Saathi. Educators can interact with the assistant in their preferred language, which responds and generates materials accordingly. Currently, the tool supports 8 languages, with flexibility to add more as needed.

The underlying intelligence powering ATL Saathi comes from Gemini, a model that creates concise instructional materials, including AI infographics, video overviews, and interactive quizzes for core curriculum modules. This helps teachers navigate complex training materials seamlessly and adopt micro-learning content efficiently.

Looking ahead, the initial rollout of ATL Sahti will involve 100 pilot schools across India. The aim is to see significant reductions in administrative load, higher efficiency, and an increased readiness among educators to help students tinker and innovate. By shifting the burden of administrative overhead and curriculum translation onto AI, educators can focus on what they do best: mentor, inspire, and guide.

The future holds promise for Indian education with ATL Saathi at its forefront. As technology continues to play a vital role in shaping educational outcomes, this tool is poised to make a significant impact.

The decline of Stack Overflow, a popular platform for software developers, has raised concerns about the future supply of high-quality training data. This issue is not just limited to one platform but may be indicative of a broader problem in the field of artificial intelligence (AI).

A recent study from the University of Auckland suggests that highly skilled contributors are increasingly disengaging from online communities as AI compresses the distinction between knowledge and AI-generated responses.

The trend is visible in Stack Overflow’s traffic, which has recorded a nearly 76% decline in monthly questions since ChatGPT’s launch in late 2022. Developers now increasingly turn to conversational AI for assistance, rather than seeking help from human experts on platforms like Stack Overflow.

This shift raises long-term questions about the future supply of high-quality training data. The departure of expert contributors from platforms like Stack Overflow may signal a bigger problem: the erosion of incentives that drive knowledge sharing among developers.

According to Dr. Kenny Ching, who led the research, AI-generated responses are becoming increasingly difficult to distinguish from those written by specialists. This phenomenon is described as ‘signal compression,’ where the value of human expertise diminishes when AI systems can generate comparable responses.

If everybody can create a good quality response or output using AI, some people may think, ‘Why should I make an effort to share my expertise and participate?’ Dr. Ching’s quote highlights the challenge facing organizations that rely on open platforms for knowledge sharing.

The implications extend beyond software development. Similar dynamics could emerge across classrooms, workplaces, research communities, and other collaborative environments where AI-generated content increasingly resembles expert work.

As the perceived value of expertise declines, organizations may find it harder to encourage meaningful knowledge sharing. This raises broader questions about AI’s own future development, particularly when it comes to training data.

Today’s large language models were trained on vast quantities of publicly available, human-generated knowledge, including content from communities such as Stack Overflow. If fewer experts continue contributing to open platforms, future training data may become increasingly fragmented or shift toward private collaboration channels.

This does not necessarily imply that future AI models will become less capable. However, it suggests that the open knowledge ecosystem that helped fuel the first generation of generative AI may be changing. As expert participation declines, maintaining high-quality public repositories of human knowledge could become as important as improving the next generation of AI models.

Financial technology giant Revolut has teamed up with leading AI platforms to bring conversational trading to its users. The integration of its crypto exchange, Revolut X, with these AI assistants lets people trade and analyze their portfolios using natural language commands.

The list of integrated tools includes Anthropic’s Claude, Google’s Gemini, OpenClaw, and Cursor. These partnerships allow users to execute trades, backtest strategies, and set up custom alerts through simple conversational prompts.

Revolut X now supports a wide range of features that make it easier for traders to analyze the market and manage their portfolios. Users can place both market and limit orders, backtest strategies across over 300 tokens available on the platform, and even receive Telegram notifications when specific price targets are hit.

The technical underpinnings of this feature rely on Anthropic’s Model Context Protocol (MCP), which was first released in November 2024. This protocol is layered on top of Revolut X’s existing trading API to enable seamless integration with the AI assistants.

Revolut has also made available a dedicated GitHub repository that provides installation support for more than 50 AI applications. This move shows the company’s commitment to making its platform accessible and user-friendly for developers as well as users.

As before, all orders require user review and approval before execution, ensuring that no trade goes through without human confirmation. Revolut explicitly disclaims liability for any errors caused by AI-driven actions – they’re not taking on extra risk just because it’s a machine making the call.

The integration of these AI assistants with Revolut X comes hot on the heels of AIR, its in-app AI assistant for general financial tasks, which launched back in April 2026. This new feature is specifically designed for traders and operates at a deeper technical level than its predecessor, enabling users to tackle more complex trading strategies.