Latest News

Claude, the artificial intelligence model developed by Anthropic, will soon start watermarking its generated content. This move is in compliance with EU rules, specifically Article 50(2) of the EU AI Act, which requires transparency in AI-generated content. The company has signed the Code of Practice on Transparency of AI-Generated Content and outlined how it plans to implement these regulations.

In a support document, Anthropic explained that models launched in the European Union (EU) after August 2, 2026, will have watermarking capabilities at launch. This includes output from various Claude platforms such as the API, Code, Cowork, and Tag. The watermarks will be embedded in generated text files and include digitally signed provenance metadata for supported file types.

The watermarking feature is not limited to Europe; it will be applied globally wherever Claude is offered. However, Anthropic notes that some platforms or features might not support certain types of watermarks. This means users may still encounter AI-generated content without visible watermarks if they don’t know what to look for.

For text files, the watermark will be integrated into the generated text itself and travel with it when copied or pasted elsewhere. For images and other file formats, Anthropic plans to use the Coalition for Content Provenance and Authenticity (C2PA) open standard. Details on recognizing these watermarks are expected in forthcoming documentation.

Existing models launched before August 2, 2026, will also receive watermarking support at a later time. However, it’s worth noting that watermarking may not be foolproof. Heavily edited or translated text might not carry detectable watermarks, and metadata can be stripped through format conversion or other means.

Anthropic’s move follows similar commitments by AI companies like OpenAI, Google, and Meta. This development aims to increase transparency in AI-generated content and help users identify the origin of such material.

Real estate technology company Inside Real Estate has announced two significant advancements in its artificial intelligence (AI) strategy: the launch of ComplianceAI, an automated compliance-review capability built in partnership with SetWave, and early support for MCP capabilities that enable direct integration between AI assistants and BoldTrail BackOffice. These moves mark a major shift towards creating intelligent back offices that can handle complex tasks efficiently.

The new autonomous compliance capability, powered by SetWave’s technology, is designed to streamline the review process of transaction files. According to data from SetWave, a substantial portion of submitted files contain errors or omissions, and reviews often take days or weeks after the file has closed. ComplianceAI aims to move this review to the moment of submission.

Compliance remains one of the most persistent operational burdens for brokerages. The introduction of ComplianceAI changes how brokerages approach compliance by shifting it from a post-factum check to an integral part of every transaction from the start. Throughout the transaction, ComplianceAI runs full compliance checks, validating required fields and signatures, flagging missing or inconsistent items, and cross-referencing the entire packet.

Agent visual compliance reports are generated automatically, with results updating in BoldTrail BackOffice without requiring a separate tool or login. This integration ensures that brokerages can catch issues before closing instead of after, reducing the risk of errors and omissions.

As brokers themselves experienced this problem firsthand, SetWave’s Co-Founder Dalton Brewer notes: ‘Partnering with Inside Real Estate lets us put that solution directly in front of the brokerages who need it most.’ This collaboration enables real estate companies to leverage AI assistants within their existing workflow, streamlining compliance checks and reducing administrative burdens.

Inside Real Estate has also confirmed early access for MCP capabilities. An open standard, MCP gives AI assistants a secure direct line into transaction and back office operations. With the broader shift towards agentic AI in 2026, Inside Real Estate worked closely with customers to develop this capability, allowing brokers, staff, and agents to surface workflows most important to them.

Key features of MCP capabilities include generating, validating, and presenting offers with embedded forms and eSignature; flagging issues before a deal is at risk; eliminating double entry by enriching transaction data automatically; strengthening the consumer experience through transaction timelines and post-close follow-ups; opening new revenue streams by tracking ancillary services; and making every agent and staff member sharper with bespoke pipeline reporting.

According to Joe Skousen, CEO of Inside Real Estate: ‘Brokerages shouldn’t have to choose between a powerful platform and compliance they can trust. ComplianceAI brings that compliance directly into the workflow our customers already use.’ The integration of MCP capabilities takes this further by enabling agents to open transactions, prepare paperwork, and get it signed just by asking – all within their existing system.

This move is part of a larger shift underway at Inside Real Estate. Their agentic AI platform, Streams Studio, connects brokerages’ entire tech stack, not just BoldTrail BackOffice. Paired together, brokers, teams, and agents can enjoy fully AI-enabled real estate operations from compliance and transactions to lead generation, marketing, and beyond.

Inside Real Estate serves nearly 400,000 top brokerages, agents, and teams as a trusted technology partner. Their branded portfolio, BoldTrail, delivers growth engines for real estate businesses with tools for lead generation, marketing automation, client relationship management, back office solutions, and best-in-class integrations.

Anthropic, the company behind the popular AI model Claude, has announced that it will be watermarking text generated by its models. This move is in response to the European Union’s (EU) new regulations on artificial intelligence, which require companies to clearly identify AI-generated or edited content.

The EU’s Transparency Code, which took effect on August 2, mandates that AI companies mark their generated or edited content in a way that can be easily identified by other systems. To comply with this regulation, Anthropic will be applying watermarks to all text generated by its models, including Claude.

All models released after August 2 will automatically include the watermarking technology, which will identify both computer-generated text and files as AI-created. For files, the company is using the C2PA open standard, a widely accepted method for verifying digital content’s authenticity.

The watermark will not only be present in new models but also extended to older ones, ensuring that all users are compliant with the EU regulations. Additionally, when users copy or paste text from an AI-generated source, the watermark will travel with it, persisting through some editing processes.

According to Anthropic’s support page, ‘the watermark is part of the text, so it will travel with the text when it’s copied and pasted elsewhere.’ However, it remains unclear how much editing users need to do to remove the watermark. We have reached out to Anthropic for clarification on this matter and will update our story accordingly.

Anthropic’s decision to implement watermarks is not unique; several other companies are also taking steps to comply with the EU regulations. Last week, AI music platform Suno announced that it would be marking tracks created on its platform after facing backlash from users and regulatory scrutiny. Newsletter service Substack has also partnered with Pangram to flag AI-generated content.

Other notable companies like Black Forest Labs, Google, Meta, Microsoft, OpenAI, and Synthesia have committed to adhering to the EU’s code, demonstrating a growing trend towards transparency in the AI industry.

MENLO SECURITY, a leader in browser security for human and agentic workforces, has expanded its Menlo Agent Runtime Security (MARS) platform with new capabilities that extend the controls enterprises use to govern their people to the AI agents working alongside them. The updated MARS platform now secures AI assistants, coding agents, and autonomous agents by sanitizing web pages and files they read, neutralizing prompt injection, and blocking data exfiltration.

The release addresses a pressing concern for businesses: securing the rapidly growing number of AI-powered tools that are being deployed in the workplace. These include browser assistants like Microsoft Copilot and Google Gemini, coding agents like Claude Code, desktop assistants like Claude Cowork, and autonomous agents that require web access. MARS is the first Browser Security Platform to strip prompt injection and hidden instructions out of what agents read and to keep them from leaking sensitive data.

Prompt injection refers to a type of attack where an AI agent reads and carries out an instruction hidden in white-on-white text, a file’s metadata, or a single human-invisible pixel. This threat is particularly concerning because it can be executed without the knowledge of even the most vigilant security teams. Mandiant’s 2026 AI Risk and Resilience report highlights prompt injection as one of the top threats to AI applications.

The growing adoption of AI agents in enterprises has created a new challenge for security teams: they must either block these agents or allow them, relying on built-in controls that were never designed for the agentic attack surface. MARS is designed to bridge this gap by providing essential runtime guardrails so security teams can confidently say yes to agentic AI instead of standing in its way.

According to Bill Robbins, Chief Executive Officer of Menlo Security, ‘Every enterprise is racing to put AI agents to work, but agents operate at machine speed without human skepticism. They live in the browser and the everyday tools our teams rely on, which is exactly where attackers are aiming.’ MARS provides a comprehensive solution for securing these AI-powered tools by running web activity in the Menlo Cloud and cleaning pages and files before they reach the agent.

The platform’s isolation-first design aligns with emerging standards, including the OWASP agentic guidelines, the NIST AI Risk Management Framework, and the EU AI Act. MARS operates at a layer that is critical for securing AI agents: everything around them, not just their models. This includes content and integrations they depend on.

Autonomous agents and coding agents run on endpoints or in the cloud, reaching the web and files through MARS. The platform cleans pages and files before an agent sees them, preventing attacks rather than detecting them after the damage is done. Five key capabilities set MARS apart: isolation and threat removal, file sanitization, adaptive data governance and protection for agents, agent authentication and attribution, and human oversight and forensics.

Isolation and threat removal involve stripping prompt injection and hidden instructions from web pages and files an agent reads. File sanitization ensures that files pulled from sources like SharePoint and OneDrive are cleaned before an agent ingests them, removing malware and hidden instructions. Adaptive data governance and protection for agents apply granular policy controls to mask sensitive information and control which websites and applications each agent can access.

Agent authentication and attribution provide token-based authentication for the agents driving secure browser sessions, enabling seamless authentication to the Menlo platform and per-agent attribution for policy and visibility. Human oversight and forensics offer security teams per-agent activity logs, session recording, and the ability to take over live agent browser sessions, backed by a tamper-proof record of content it saw and acted on.

Because MARS runs on the same Browser Security Platform Menlo already uses to govern human employees, security teams apply one policy framework across both people and agents. This streamlines their work and reduces complexity. The platform’s isolation-first design also aligns with emerging standards for securing AI-powered tools in enterprises.

The decision to build MARS follows more than 70 conversations with customers over the past three months, which surfaced concerns about assistants with broad access to email and files, low-code agents inheriting employee credentials, coding agents exposing source code, and autonomous agents exposed on the open web. Menlo’s solution addresses these pressing issues head-on by providing a comprehensive platform for securing AI-powered tools in enterprises.

AI assistants are increasingly being used in businesses as data analysis tools to streamline operations and improve efficiency. However, their growing adoption has also created new security challenges that must be addressed. MARS is designed to provide the essential runtime guardrails so security teams can confidently say yes to agentic AI instead of standing in its way.

The platform’s capabilities are crucial for securing AI-powered tools in enterprises because they address a critical gap in current security solutions: protecting everything around AI agents, not just their models. This includes content and integrations they depend on. By providing a comprehensive solution for securing these tools, MARS helps businesses to confidently deploy AI assistants and coding agents without compromising their security posture.

Trimble has introduced a new artificial intelligence (AI) solution designed to automate manual workflows and streamline operations for global supply chains. The company’s latest innovation is called Trimble Arc Agent, which aims to simplify tasks and reduce reliance on multiple agents.

The single AI agent can handle complex tasks with ease, moving beyond the limitations of multi-agent setups that often hinder productivity in logistics teams. By automating office work and eliminating manual processes, Trimble Arc Agent enables organizations to focus on high-level strategy where human decision-making is crucial.

According to a recent survey conducted by Trimble and FreightWaves, 70% of transportation and logistics companies struggle with manual, repetitive tasks as their leading pain point. These tasks not only consume staff time but also stifle productivity.

The solution connects directly to existing transportation management systems (TMS) and third-party applications like Gmail and Outlook. By securely extracting and validating data from unstructured channels such as emails, PDFs, and spreadsheets, Trimble Arc Agent allows teams to turn real-world inputs into actionable system-of-record data.

Trimble Arc Agent is built with core agentic safety features that minimize the risk of AI hallucinations. This ensures organizations can drive faster throughput and reduce operational costs without compromising service levels or margins. The solution addresses industry concerns regarding AI reliability, making automation a viable option for companies looking to improve efficiency.

The skills catalog at the heart of Trimble Arc Agent is designed with governance controls and risk review workflows in mind. By combining a library of ready-to-use skills with custom skill-building capabilities, organizations can address their biggest challenges through simple natural language interactions.

Example skills include data entry, market insights, customer support, and personal assistant functions. These features allow teams to automate tasks such as order entry and contract intake, extract full truckload rate intelligence, provide contextual troubleshooting assistance, or consolidate daily workflows into a single overview.

The solution is available for Trimble customers worldwide on a single-tier subscription model with no seat limits or feature gates. Users receive 10 hours of agent working time (with additional overage hours available) and automatic access to all new skills and features as they launch.

Trimble Arc Agent’s design prioritizes safety, reliability, and explainability. The solution is built on enterprise-grade AI guardrails and human-in-the-loop controls that ensure AI-driven decisions are auditable and supportable. This approach addresses industry concerns regarding AI reliability so automation does not come at the cost of service commitments or compliance.

The Trimble Arc Agent skills catalog will continue to expand as new skills are added over time, allowing organizations to adapt their workflows according to changing needs. The solution’s intuitive interface enables users to easily build custom skills without requiring engineering resources.

Trimble is a global technology company that connects the physical and digital worlds, transforming the ways work gets done. With relentless innovation in precise positioning, modeling, and data analytics, Trimble enables essential industries including construction, geospatial, and transportation.

The introduction of Trimble Arc Agent marks another step forward for Trimble’s efforts to streamline operations and improve efficiency across various sectors. By providing a comprehensive solution that addresses the challenges faced by logistics teams, Trimble is helping organizations automate manual workflows and unlock new levels of productivity.

Reddit has long relied on volunteer moderators to keep its communities free of abuse and other unwanted content. However, the company is now exploring a new approach: using artificial intelligence (AI) bots to help moderate subreddits. The move comes as AI automation continues to transform industries and potentially displace human workers.

The introduction of AI moderation tools, dubbed ‘Rules Hub,’ aims to streamline community management by leveraging large language models to enforce rules set by users. According to Reddit’s blog post, this new system will be able to better handle nuances in language and edge cases compared to existing automation tools like Automod.

However, the effectiveness of these AI mods remains uncertain. Some users have expressed concerns that relying on AI bots may not address the root issues driving abuse and spam on the platform. As one user noted, ‘I do not expect this to work well with [lyrics as comments]… I doubt AI will be able to grasp the nuance of the lyrics of someone like Lee Scott or Jam Baxter.’

Reddit’s decision to introduce AI moderation tools has sparked a mix of reactions on the platform. While some users see it as an opportunity to improve community management, others are skeptical about relying on machines to handle complex tasks that require human judgment and empathy.

The company claims that its new system will make Reddit more accessible and straightforward while maintaining what makes the platform special – namely, user-generated content. However, this goal may be at odds with the increasing presence of AI-generated content on the site, which has led some users to question whether the ‘human’ aspect of Reddit is being compromised.

Reddit’s reliance on volunteer moderators has long been a contentious issue, with varying degrees of success and reputation among community members. The introduction of AI moderation tools raises questions about the future role of human moderators and whether they will be replaced by machines in the near future.

The company hopes that its new system will eventually replace many of the enforcement workflows that communities rely on from Automod, an existing automation tool that has been a thorn in the sides of users. However, it remains to be seen whether AI bots can effectively handle the complexities of community management and address the root issues driving abuse and spam on the platform.

Reddit’s decision to introduce AI moderation tools is part of its ongoing efforts to combat AI-generated content and bot spam on the site. The company has faced criticism in the past for allowing AI companies to access Reddit data without paying, which led to an outright rebellion by users when new API restrictions were introduced.

The introduction of Rules Hub marks a significant shift towards relying on AI tools for community management, but its effectiveness remains uncertain. As one user noted, ‘Before long, Reddit will be mostly AI bot accounts, moderated by AI bots, discussing mostly AI-generated slop.’ This prediction highlights the concerns that some users have about the future direction of the platform and whether it will remain a bastion of human-generated content.

The use of large language models in Rules Hub allows for more nuanced evaluation of posts and comments compared to existing automation tools. However, this advantage may not be enough to address the complex issues driving abuse and spam on the site.

Teams are rapidly integrating Claude Enterprise, an artificial intelligence (AI) assistant, into their workflows. However, this integration also creates a new attack surface for potential security threats. The prompt used to interact with the AI assistant is now exposed and vulnerable to manipulation by malicious actors.

The traditional approach to data loss prevention focuses on identifying sensitive data leaving the organization’s systems. However, this method does not account for the manipulation of the model itself or its tools. A jailbreak can attempt to bypass the assistant’s guardrails, while a prompt injection hidden in a shared file can hijack an agent and turn it against the user.

Anthropic recently introduced inference hooks, which enable organizations to send each governed prompt to an AI security service before inference begins. When configured, this hook allows Cisco AI Defense to inspect the prompt for artificial intelligence threats and return a verdict of allow or deny. This integration is crucial in protecting against potential attacks on the model and its tools.

The focus on AI-first defense by Cisco sets it apart from traditional data loss prevention methods. While these methods are designed to identify sensitive data, Cisco’s approach reads intent and catches prompt injection and jailbreaks that target the model, its tools, and agents acting on its behalf. This evaluation also includes agent activity in the conversation transcript.

When Claude Code or Cowork calls a tool, including those connected over the Model Context Protocol (MCP), the transcript can include the tool call and its result. Cisco AI Defense runs both privacy and manipulation guardrails on this transcript: it flags sensitive data as traditional data loss prevention would, plus identifies injection and jailbreak attempts that data loss prevention was never designed to detect.

This reflects how we think about the entire problem of securing AI assistants in business environments. The Claude Enterprise integration is one runtime enforcement point within a broader platform. Cisco AI Defense secures the full AI lifecycle by discovering and inventorying AI assets, assessing their risk through validation, red-teaming, and supply chain and model scanning, and protecting them at runtime with guardrails like this one.

The result of this integration is real-time protection delivered through the hook, an additional layer for comprehensive AI coverage that adds enforcement without requiring infrastructure changes to the user’s workflow. This gives security leaders a direct enforcement point for AI policy across governed Claude Enterprise requests. Enterprise users can maintain their fast and native experience while using Claude.

For teams relying on data analysis tools like Claude Enterprise, this integration provides an essential layer of protection against potential threats. By leveraging Cisco AI Defense, organizations can ensure the secure use of these powerful AI assistants in their business environments.

Cisco AI Defense plugs into Anthropic’s inference hooks to provide inline protection for every governed prompt: The prompt is sent for inspection; Claude Enterprise sends each governed prompt across Claude, Code, and Cowork to Cisco AI Defense through the hook before model execution. This ensures that potential threats are caught early in the process.

The request is cryptographically verified using a one-time signing secret. Cisco AI Defense checks this signature and confirms the request’s authenticity before inspecting it for potential threats. The conversation transcript, including tool calls and their results, is then evaluated against the organization’s runtime policy for prompt injection, jailbreaks, tool exploitation, and sensitive data.

A verdict returns before inference begins: allow or deny. Safe prompts continue through to model execution; a malicious prompt is blocked, preventing potential harm.

Cisco AI Defense inspects both prompts and responses in this integration. This uses Anthropic’s inference hook, which currently fires on each governed prompt before the model runs. As Anthropic extends the hook to responses, response-side enforcement can use the same integration path. Over time, we expect this enforcement to become a native part of the Cisco AI Defense Inspect API.

This integration is available today with Claude Enterprise and inference hooks in beta. Users can try it out by visiting our developer portal’s playground; enter a prompt and watch the live inspection and verdict. Schedule time with our team to see how Cisco AI Defense stands up against AI threats.

A new tool from Anthropic is set to make it harder for people to pass off AI-generated content as their own. The company has announced that its Claude models will now embed an imperceptible watermark into text generated by the AI, making it easier to detect and investigate cases of plagiarism.

The watermark doesn’t change the meaning or readability of the original text but can be detected when copied and pasted. It may also persist through some editing processes, according to Anthropic. This move is part of the company’s efforts to increase transparency in line with the European Union AI Act.

A watermark is essentially a digital signature that proves ownership and authenticity of a piece of work. In this case, it will be embedded into text generated by Claude worldwide, including when accessed through cloud providers. Anthropic has also committed to providing third parties with tools to detect these watermarks.

The introduction of AI watermarking could have significant implications for the publishing industry, which has been grappling with concerns over AI-generated content. Last month, a book agent pulled support for a novel after allegations emerged that its author may have used AI to write it. The writer denied any involvement, but the incident highlights the need for greater transparency and accountability.

Anthropic’s move is not an isolated one - Google DeepMind has also been watermarking text generated by its Gemini app using SynthID technology since 2024. This development could give publishers, schools, and universities another tool to investigate cases of plagiarism and make it harder for people to pass off AI-generated writing as their own.

However, there are still potential loopholes in the system. Anthropic notes that heavy editing or paraphrasing can potentially remove the watermark, making it undetectable. Additionally, finding a watermark does not necessarily prove that Claude was the original author of the content, since even proofreading or translating with its AI can leave behind a mark.

The rollout marks a significant step towards greater transparency and accountability in the use of AI-generated content. As more companies begin to adopt similar measures, it will be interesting to see how this development impacts the publishing industry and beyond.

Athropic is making changes to its language model, Claude. The company’s adjustments are a response to new regulations from the European Union.

The push for transparency in AI-generated content comes from Article 50 of the EU AI Act. This article requires companies like Anthropic to be more open about their use of this technology. As part of complying with these regulations, Anthropic has signed up to the European Union’s Code of Practice covering AI-generated content.

Starting August 2, 2026, new Claude models will include watermarks at the model level. These watermarks appear in all generated text - regardless of how it’s accessed. This includes major platforms like AWS and Anthropic’s own cloud customers such as Google Cloud and Microsoft Foundry among others.

The watermarks are one aspect of a broader plan by Athropic to increase transparency around AI-generated content. The company is also working on a system for supported files that will include signed provenance metadata based on the C2PA standard.

This framework allows users to record information about digital content, enabling verification whether a file has been processed or altered in any way. For instance, it’s particularly important when verifying AI-generated images like those created using Claude for artistic purposes - which can be used for various applications including advertising and educational materials.

Anthropic cautions that text watermarks nor signed provenance metadata should not be considered definitive proof of authorship. The company notes its models can process original material from people, resulting in the output carrying a mark indicating it was processed by Claude even if not generated entirely by humans.

The technical challenges surrounding these signals are significant with text watermarks able to travel and file metadata potentially disappearing during format conversions or re-saving. Athropic is working on providing more detailed documentation as its marking systems mature.

For businesses that rely heavily on AI tools for data analysis, the addition of watermarks marks a crucial step towards greater transparency in AI-generated material - something that can make a big difference when assessing credibility. By providing machine-readable identity, Anthropic’s watermarks offer another way to verify whether content has been processed by an AI model.

Older models won’t get these new features right away but Athropic is working on adding marking capabilities during the transition period. This will ensure all users can benefit from increased transparency in generated content - something that could help platforms more easily identify when text was processed using Claude.

The changes may also impact how customers use AI-generated images which are used for various purposes including advertising or educational materials. By providing a way to verify whether an image was created with the model, businesses can have greater confidence in their content and avoid potential issues related to mislabeled material.

As new technologies emerge like AI tools that generate data analysis solutions - transparency becomes increasingly important. With watermarks and signed provenance metadata Anthropic is taking steps to address these concerns by providing a way for platforms and organizations more easily identify when a piece of text has been processed using its language model.

A new tool has been developed to identify the sources of fake videos, a growing concern in today’s digital landscape. The innovation comes as artificial intelligence systems are rapidly improving and synthetic videos are increasingly being used in entertainment, advertising, education, social media, and politics.

The AI-generated video problem is not just about determining whether a video is real or fake; it’s also about understanding where the content originated from. A team of researchers at UC Riverside has developed a tool called SAGA (Source Attribution of Generative AI Videos) that can identify which AI system created a particular video.

SAGA works by analyzing visual patterns within individual frames and temporal relationships across entire video sequences. Unlike still images, videos contain motion and temporal information - how visual elements change from one frame to the next over time. Different AI video generators produce subtle patterns in those changes, creating distinctive artifacts that can be detected.

The researchers used a technique called Temporal Attention Signatures (T-Sigs) to visualize these unique patterns associated with different video generators. By averaging patterns across many videos produced by the same AI system, SAGA generates a characteristic profile that helps distinguish one generator from another.

SAGA was tested using public datasets containing videos generated by 19 different AI video generators, including text-to-video systems and image-to-video systems. The tool can determine whether a video is real or synthetic, identify whether it was created from text or images, and even identify the development team behind a model.

The ability to identify the source of AI-generated content could help digital forensic investigators track misinformation campaigns, assist regulators in enforcing transparency requirements, and aid technology companies in understanding how synthetic media spreads online. This is particularly important when it comes to free AI video generators that are increasingly being used by businesses for advertising and other purposes.

According to Rohit Kundu, the lead researcher on the project, ‘The patterns are like fingerprints that the generative model leaves behind.’ The researchers’ goal was to find out if these signatures were distinct amongst different generators. Their findings show that yes, there are distinct fingerprints left by each AI system.

The development of SAGA is part of an ongoing effort to make generative AI safer as increasingly sophisticated AI video tools become widely available. As Kundu noted, ‘It’s a cat-and-mouse game for sure.’ The researchers’ work aims to stay ahead of the curve and provide a comprehensive solution to identifying the sources of fake videos.

The research is described in the paper ‘SAGA: Source Attribution of Generative AI Videos,’ which includes contributions from Kundu, Professor Amit Roy-Chowdhury, and other collaborators. Their current work builds on previous research by the same team that developed an artificial intelligence model to detect video tampering.

Finance has undergone a significant transformation in recent years, shifting from a static function to a real-time one. This change is not just about closing the books faster or refreshing forecasts more often; it’s about seeing the business as it changes and helping leaders act sooner. A finance team that can provide this level of insight can give the company more time to shape what happens next.

When I joined OpenAI two years ago, we had a small finance team supporting a rapidly growing company. We needed to build our function from scratch and make AI fundamental to how we work, make decisions, and support the business. This meant redesigning our processes around artificial intelligence tools that could help us analyze data more efficiently.

One of the key challenges we faced was manual, recurring work: finding information, explaining what changed, and assembling inputs for a decision. We had access to advanced AI tools but were still learning how to use them effectively. So, we set two ambitious goals: achieving a zero-day close and automated, continuously updated forecasting.

A zero-day close means giving leaders a real-time view of the company’s financial position, reconciled and traceable. Continuous forecasting builds on this foundation by showing how the business is changing, what could happen next, and which decisions can alter the outcome. We’re still working toward these goals, but our efforts have already changed how we operate.

Our team has moved beyond static spreadsheets, manual searches for supporting records, and presentations to live tools built on the full context and data of the business. Finance professionals now have the ability to build their own decision-making infrastructure using AI tools like ChatGPT Work and Codex.

The real promise of an AI-native finance function is a team that understands what’s happening as it happens, helps leaders see choices ahead, and gives the company more time to act while outcomes can still change. To get there, we need to redesign work around decisions that matter, give people room to experiment, build clear accountability into every workflow, and measure AI’s dependable work.

One of our key takeaways is the importance of broad access to AI tools. People need freedom to explore these capabilities in their own work context. We paired this with structured experimentation around real problems, which led to some innovative solutions like IR-GPT, a custom GPT for investor relations.

Our hackathon turned AI from an abstract capability into a working tool. In one day, people could identify recurring tasks, build solutions, test them with colleagues, and improve them. The use cases came from those closest to the work while technical experts helped move faster.

For CFOs, our lesson is simple: you need bottom-up experimentation combined with top-down strategy. Put secure AI in people’s hands and let those close to the work identify better ways of getting things done. At the same time, focus leadership attention on changes that will matter most to the business.

The advent of artificial intelligence has sparked intense debate about its potential impact on humanity. While some see it as a powerful tool for driving societal progress, others fear that an uncontrolled AI could lead to catastrophic consequences. To shed light on this complex issue, we spoke with Robert J. Marks II, a renowned expert in the field.

Dr. Marks is a distinguished professor of engineering at Baylor University and has spent over three decades researching artificial intelligence. His work has been supported by prominent organizations such as NASA, the National Institutes of Health, and Microsoft. He has also consulted for Boeing and authored numerous books on AI-related topics.

As Director of the Walter Bradley Center for Natural and Artificial Intelligence, Dr. Marks is well-positioned to provide a comprehensive understanding of both the benefits and potential perils of AI. His latest book, Non-Computable You: What You Do That Artificial Intelligence Never Will, explores the limitations of artificial intelligence in relation to human capabilities.

Some proponents argue that AI will lead to significant improvements in areas such as medicine, economics, and knowledge acquisition. However, Dr. Marks cautions that these benefits must be carefully weighed against potential risks, including loss of liberty and societal destruction.

The possibility of an out-of-control AI is a pressing concern for many experts. While some may see it as a distant threat, others believe it could have devastating consequences if not addressed promptly. As we continue to develop and integrate AI into our lives, it’s essential that we consider both the benefits and risks associated with this technology.

Dr. Marks’ extensive experience in AI research has given him a unique perspective on its potential impact. He emphasizes the need for careful consideration of the long-term consequences of developing increasingly sophisticated artificial intelligence systems.

A security firm has discovered that an AI assistant used by businesses can be compromised with hidden instructions in uploaded files. PromptArmor found that the Rovo AI assistant, made by Atlassian, can be steered to exfiltrate data without human approval when a poisoned file is uploaded. This vulnerability persists even if web search functionality for Rovo is disabled.

The technique exploits how AI models process text, making it difficult to distinguish between legitimate user input and hidden instructions. An attacker can embed malicious commands in a PDF document that the model will interpret as genuine requests. For instance, an instruction written in transparent color and a font size of 1 pixel may be invisible to human eyes but still recognizable by the AI.

This type of attack is known as prompt injection, where someone inserts unauthorized instructions into content being processed by an AI system. In this case, Rovo’s job is to read documents and act on them accordingly. If a hidden line in a document instructs it to send confidential tickets to an attacker-controlled URL, the model will comply without raising any alarms.

PromptArmor notes that even if organizations disable web search for Rovo, the vulnerability remains because the tool used to open search results is still active. This means that disabling one feature does not necessarily eliminate the risk of data exfiltration.

The security firm tested several AI agents built on GPT-5 and Gemini and found that more than 79% failed to resist prompt injection in direct tests. The Rovo vulnerability demonstrates how this pattern can repeat with agents designed for reading and acting, pointing them in the wrong direction.

Atlassian received PromptArmor’s report on May 23 but has since gone silent despite multiple follow-ups from the security firm. As a result, Rovo remains vulnerable to prompt injection attacks, leaving businesses that rely on it exposed to potential data breaches.

A significant development in the field of mathematics has emerged from an unexpected quarter: AI model Claude. In a recent challenge, Claude was tasked with tackling one of the most famous unsolved problems in mathematics – the Riemann hypothesis. Although it didn’t succeed in resolving the problem itself, its attempt led to an important breakthrough on a related issue.

Researchers at Anthropic have been studying Claude’s capabilities and recently gave it an unreasonable challenge: take a stab at the Riemann hypothesis. This task has been open since 1859, with a million-dollar bounty still unclaimed. The team was aware that Claude might not succeed in resolving the problem itself, but they were eager to see how it would approach this complex issue.

During its attempt, Claude unexpectedly made significant strides on a related problem – improving the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the Riemann hypothesis. This improvement is substantial: from 41.6% to 67.2%. The new result draws heavily on prior research by mathematicians over several decades, combining results from Baluyot, Goldston, Suriajaya, and Turnage-Butterbaugh with Bombieri’s work.

The Riemann zeta function describes the distribution of prime numbers – each zero that contributes to this sequence provides finer detail. The Riemann hypothesis states that these zeros exist along a certain vertical line. This conjecture has become one of the most consequential in mathematics, as many results assume it to provide randomness in primes.

No one has yet been able to prove or disprove the Riemann hypothesis, but mathematicians have made progress in related areas. One such direction concerns quantifying a minimum proportion of zeros that are on the line – over time, this known constant proportion has gradually increased to 41.6%. Another area focuses on the distribution of zeros along the line.

Claude’s result shows how AI models can extend and build upon mathematicians’ ideas in new ways. Even though it couldn’t resolve the Riemann hypothesis itself, its breakthrough emerged as an unintended byproduct of that original request. This development highlights the potential for AI to make meaningful contributions to mathematical research – a field often considered one of human ingenuity.

The team at Anthropic is grateful for the input and expertise provided by Brian Conrey and Dan Goldston, two experts in this area who generously examined Claude’s paper on short notice. The full technical explanation of Claude’s finding is available in its paper, while an informal note concisely stating Claude’s proof has been produced by Levent Alpöge and Ralph Furman.

Claude arrived at the new lower bound over two sessions using a total of 31 million output tokens in Claude Code. This achievement demonstrates how AI models can tackle complex mathematical problems with speed and efficiency – an area where human mathematicians often struggle to make significant progress.

The process behind this breakthrough is fascinating, involving multiple subagents working together to run thousands of numerical checks against known zeta zeros. The team at Anthropic used a combination of encouragement and technical guidance to help Claude overcome initial skepticism about its ability to make meaningful contributions.

Claude’s work also highlights the potential for AI models like itself to write formal proofs, recommend human validation, and even produce Lean formalizations – as seen in its collaboration with Eric Easley. This development underscores the importance of exploring how AI can assist mathematicians in their research efforts.

Generative artificial intelligence has brought about an increased awareness of its underlying technologies, including natural language processing and large language models (LLMs). This growing understanding creates opportunities for crucial conversations about how these components impact the effectiveness and suitability of AI solutions in mission-critical contexts. One key area to explore is LLMs themselves.

A large language model is essentially a neural network trained on vast amounts of text, images, and video data. Its purpose is to recognize patterns in language, predict words, and generate coherent sequences that resemble natural language. This enables machines to answer questions, perform translations, summarize human-generated text, or optimize it for specific purposes.

However, just like specialized professionals such as doctors and attorneys require tailored expertise, LLMs must be adapted for their intended use case. A geospatial model designed to detect tumors in X-rays is not suitable for a mission that requires domain-specific knowledge. In the context of GDIT’s work with its partners, what’s needed is an understanding of how these models can be optimized for specific tasks and environments.

The complexity of LLMs lies in their ability to handle vast amounts of data. Models like ChatGPT have billions of parameters; by streamlining this process, organizations can operate more efficiently, access their AI tools from remote locations (such as the edge), and achieve more accurate results.

When implementing AI and LLMs for mission-critical applications, it’s essential to consider several key questions. First, will you maintain control over your data? This includes in-house information like emails, records, file shares, document repositories, and internal documents – all of which contain language unique to the specific mission space.

Retaining full ownership allows organizations to enforce governance, privacy, and compliance policies. It’s crucial for teams to be able to share their data with LLMs securely and ensure it remains protected throughout the process. This involves not only sharing but also adapting to new context, contributors, and concepts that emerge over time.

Another critical consideration is continuous improvement. Can you adapt your models as needed? How does the LLM incorporate new information or changing context, and how are these changes preserved within the model? These factors directly impact sustained accuracy and trustworthiness of the AI solution.

It’s also vital to assess whether an LLM truly understands the mission space and its requirements. What data was it initially trained on, and how did it perform in that environment? This evaluation will help determine if the LLM is genuinely suited for your specific needs or merely a facsimile.

Lastly, teams must evaluate models against potential risks such as manipulation, corruption, and bias within their supply chain processes. They should also be aware of methods to detect, mitigate, and prevent data poisoning through robust quality control measures. Starting this dialogue early can help avoid costly issues down the line.

SFA athletics has made a significant move in the world of sports storytelling by partnering with FanWord, an AI-powered platform used by over 215 schools and conferences across the United States and Canada. The partnership aims to ‘humanize’ box scores and stats by focusing on the people behind them rather than just numbers themselves.

The multi-year deal gives SFA’s athletic communications team access to FanWord Assist, a suite of tools designed to turn box scores, stats, and postgame interviews into ready-to-edit game recaps, player bios, and award nominations. This AI tool is not only used by Stephen F. Austin but also by other prominent institutions such as Florida State, UConn, and SMU.

According to Jacob Pumphrey, SFA’s assistant athletic director for communications, the goal of this partnership is simple: tell better stories faster while still maintaining a human touch in sports reporting. ‘The human element is not going to be lost,’ he emphasized. ‘We still have to be good crafters of the written word. We still have to know how to storytell.’

Pumphrey explained that FanWord Assist will help his three-person communications office work more efficiently, freeing up time to pursue deeper and more meaningful athlete stories. He noted that this partnership is not a replacement for human journalists but rather an aid in streamlining the process of creating game recaps.

The program costs between $250 and $350 per sport, which Pumphrey believes is roughly an eighth of what a full-time staff position would cost. This makes FanWord Assist an attractive option for athletic departments looking to cut costs without sacrificing quality content.

What sets FanWord apart from other AI tools is its focus on the needs of people working in athletic communications. ‘The one thing that kind of set FanWord apart from just your normal AI is the fact that it’s a company that employs people who have actually worked in these positions,’ Pumphrey said. This understanding of the industry allows FanWord to craft tools tailored specifically to the challenges faced by those in this field.

One feature that Pumphrey is particularly excited about is the platform’s first-person storytelling tool, designed to help tell athlete stories even when schedules make sit-down interviews nearly impossible. ‘Student athletes’ schedules are very busy,’ he noted. ‘It’s tough to maybe find a 30-minute window sometimes to sit down and talk with them.’ This tool allows for text answers, audio answers, or video answers from student-athletes at their convenience.

Not everyone is convinced that this partnership is the right move, however. Kylie Chandler, a junior at SFA and writer for The Pine Log student newspaper, expressed disappointment over the decision to use AI in game recaps. ‘I feel like it’d be so much more meaningful if we used this opportunity to showcase our students’ abilities,’ she said.

Chandler also raised concerns about mixed messaging on campus regarding AI usage. She noted that while SFA is moving forward with AI-powered tools, students are discouraged from using AI in their coursework. Pumphrey pushed back against the idea that the partnership signals a lack of confidence in student journalists.

‘This isn’t a move because we don’t want to rely on our journalism department or we don’t feel that students can get the job done,’ he said. ‘This is just more so to help the current full-time employees in the office do their job a little bit more efficiently.’

The door remains open for student workers, with Pumphrey encouraging any interested students to come forward and discuss potential opportunities. The athletics department says it is actively looking for student workers and interns heading into the fall season.

Ultimately, this partnership marks a significant step towards ‘humanizing AI’ in sports storytelling, but its impact will depend on how well SFA’s athletic communications team can balance efficiency with the need to tell meaningful stories about athletes.

OpenAI is expanding its Daybreak platform, which provides access to advanced cyber capabilities for trusted defenders. The move comes as the cybersecurity landscape becomes increasingly complex, with threat actors leveraging AI to conduct attacks at unprecedented speed and scale.

The company’s latest model, GPT-5.6-Cyber, is designed to improve performance on specialized cybersecurity tasks such as finding zero-day vulnerabilities and developing exploit chains. It is available through Daybreak Red access, which provides more advanced capabilities than the standard Daybreak Blue tier.

GPT-5.6-Sol, a general-purpose model, has been shown to deliver state-of-the-art performance on cybersecurity tasks in production environments. However, it can block legitimate defensive work due to system-level safeguards designed to prevent misuse. Daybreak Blue access removes these guardrails, allowing defenders to get more out of the model in real-world security tasks.

Even with reduced safeguarding, GPT-5.6-Sol may still refuse to comply with certain highly dual-use cybersecurity prompts. To address this issue, OpenAI trained GPT-5.6-Cyber to further reduce refusals and improve performance on specific tasks. The model is designed to help trusted defenders conduct legitimate security activities.

To measure the reduced rate of refusals provided by GPT-5.6-Cyber through Daybreak Red access, OpenAI created an internal evaluation called Advanced Cybersecurity Completion Rate. This metric measures how often models will respond to requests involving exploit-chain development and other advanced cybersecurity scenarios. According to this evaluation, GPT-5.6-Cyber completes 95% of these requests compared to just 1.5% for GPT-5.6-Sol.

GPT-5.6-Cyber outperforms both GPT-5.6-Sol and the previous model, GPT-5.5-Cyber, on several specialized cybersecurity tasks. On ExploitGym, which evaluates an agent’s ability to turn known vulnerabilities into working exploits in controlled environments, GPT-5.6-Cyber shows significant improvement over its predecessors.

Another area where GPT-5.6-Cyber excels is finding and accurately calibrating the severity of novel zero-day vulnerabilities. OpenAI created an internal evaluation dataset that provides models with a current release of an open-source repository, asking them to generate proof-of-concept exploits alongside a technical write-up of their findings.

GPT-5.6-Cyber also shows improvement on Vulnerability Discovery and Report Writing, which evaluates an agent’s ability to find vulnerabilities in a repo with known vulnerabilities. While GPT-5.6-Sol performs better than GPT-5.6-Cyber on this evaluation, the latter model still outperforms its predecessor.

OpenAI has also used GPT-5.6-Cyber to identify high-severity issues in various software projects, including popular databases and mobile phones. The company is working closely with Daybreak partners and members of the open-source community to disclose and remediate these vulnerabilities.

The Preparedness Framework assesses a model’s cybersecurity capability as High for GPT-5.6-Sol and similarly reaches the High threshold but not Critical for GPT-5.6-Cyber. The latter model improved over its predecessor on some specialized cyber tasks, but not sufficiently to reach the Critical threshold.

OpenAI is taking additional steps to enable safer use of cyber models through identity verification, account security, monitoring, approved-use restrictions, and legal attestations. The company also encourages Daybreak customers using Codex to switch from full-access mode to auto-review mode, which evaluates actions requiring elevated permissions before execution.

The company requires all individual accounts in Daybreak to adopt hardware security keys beginning September 1, 2026. OpenAI is actively working on additional security measures, including improved monitoring, and prioritizing alignment training and testing for upcoming releases.

A new artificial intelligence institute has opened at North Carolina Central University, marking the first of its kind at a Historically Black College and University. The Institute for Artificial Intelligence and Emerging Research is designed to provide students with hands-on experience in developing and using AI tools for businesses, as well as understanding their potential biases and limitations.

The university’s campus now features several state-of-the-art research and training spaces, including an AI literacy studio, a robotics and emerging tech lab, and an extended reality experimentation sandbox. These facilities will serve not only students but also faculty members who can leverage seed grant funding for AI-related projects.

Founding director Siobahn Day Grady emphasized that the institute’s mission is to make AI accessible to everyone on campus, including those outside of computer science departments. ‘AI used to be something specifically just for computer scientists,’ she said, ‘but with AI tools now it’s accessible to everyone.’

The university has already seen significant impact from its efforts, with over 2,500 people affected by the institute’s work in the past year and a half, including students who have taken introductory courses on AI. Starting this fall, NC Central will offer an AI minor for students interested in exploring the subject further.

Faculty members are also being encouraged to engage with AI research through seed grant funding opportunities. However, some concerns remain about the potential effects of AI on academic integrity and critical thinking skills. A recent study by the American Association of Colleges and Universities found that faculty worry about these issues.

In response to these worries, Chancellor Karrie Dixon emphasized the importance of guiding students in responsible AI usage. ‘Technology moves faster than trust,’ she said. ‘We have to educate our campus community on how to use AI in a way that’s both effective and ethical.’

The institute was made possible through a $7 million investment from NC Central University, including federal funds and a $1 million contribution from Google. On Monday, the university hosted a ribbon-cutting ceremony for the facility, attended by local government leaders such as Congresswoman Valerie Foushee.

Foushee highlighted the significance of this initiative in preparing students for high-demand careers and expanding access to emerging technologies for underrepresented communities. ‘Our future cannot be built by just a few tech billionaires and CEOs,’ she said. ‘This institute matters because it will prepare students, strengthen research partnerships, support entrepreneurs, and expand access to emerging technologies.’

A leading technology company, PTC, has introduced a new capability called Arena Connect to simplify the management of product development and quality processes. This innovative tool is designed to automate workflows across various systems used in the product lifecycle, including PLM (Product Lifecycle Management), QMS (Quality Management System), ERP (Enterprise Resource Planning), MES (Manufacturing Execution System), collaboration tools, and other enterprise systems.

According to PTC, Arena Connect addresses a common challenge faced by manufacturers: keeping data synchronized across multiple systems. As companies grow, they often rely on numerous systems for managing product development, quality, manufacturing, and business operations. However, these systems are frequently disconnected, requiring manual effort or custom integrations to maintain data consistency.

Arena Connect provides a centralized platform for connecting applications, automating data movement, and ensuring that critical business processes run smoothly across various systems. This means that engineering change orders and quality processes can be synchronized automatically between Arena and Jira, with issues created and updated as workflows progress.

The goal of Arena Connect is to enable manufacturers to connect their existing systems without creating a new system. As David Katzman, general manager of Onshape and Arena at PTC, explained: ‘Manufacturers don’t need another system to manage; they need an easier way to connect the systems they already have.’ With Arena Connect, customers can automate workflows and keep information moving between teams without the complexity associated with custom integrations.

A key feature of Arena Connect is its visual workflow builder. This tool allows users to design, build, and test workflows using no-code templates or full custom-code integrations for systems without pre-built connectors. Additionally, a connection library and workflow templates are available for creating integrations using 200+ pre-built connectors, as well as ready-to-use workflow templates for popular systems like Jira, Microsoft Teams, and Slack.

Another important aspect of Arena Connect is its ability to automate real-time and scheduled business processes using webhooks, the Arena Events Engine, and custom action triggers. This ensures that workflows are executed efficiently and accurately, reducing manual effort and minimizing errors.

Arena Connect also includes an administration and governance framework for monitoring workflow executions, logs, alerts, and error management through a centralized dashboard. This feature enables organizations to track and manage their workflows effectively, ensuring compliance with regulatory requirements and industry standards.

PTC has further strengthened supply chain resiliency in this release by introducing new features such as the Arena AI Assistant, Arena SCI (Supply Chain Insights), and Onshape-Arena Connection. These tools provide enhanced capabilities for managing product data, including AI-powered item redlining and file insights to summarize product changes, compare revisions, and find information across documents using natural language.

The introduction of Arena Connect is part of PTC’s broader vision for the Intelligent Product Lifecycle: enabling manufacturers and product companies to build a product data foundation, extend its value across their enterprise, and accelerate AI-driven transformation. By leveraging product data more effectively, companies can bring higher-quality products to market faster, better manage product complexity, meet regulatory standards, and achieve other business objectives.

PTC’s Arena Connect is now available as part of the 2026.2 release. To learn more about this capability and its benefits for manufacturers, readers are encouraged to explore further resources provided by PTC.

Artificial intelligence is rapidly advancing, but Alaska’s policy framework has failed to keep pace. While other states have made significant strides in regulating AI, the Last Frontier remains largely unprepared for its implications. This disparity becomes apparent when examining various aspects of AI governance, from chatbots and algorithmic bias to disinformation and deepfakes.

The use of companion chatbots is a prime example of where Alaska falls short. These digital assistants are increasingly prevalent in everyday life, often interacting with children and vulnerable individuals. However, unlike many other states, Alaska has no laws requiring these chatbots to disclose their non-human nature or take steps to prevent harm. This lack of regulation leaves users exposed to potential risks.

House Bill 47, which passed this June as part of an omnibus crime bill, marked a narrow step forward for AI governance in Alaska. The legislation criminalized the creation and distribution of AI-generated child sexual abuse material and deepfake harassment. While this is a positive development, it represents only a small fraction of the comprehensive framework needed to address AI’s complexities.

Algorithmic bias is another area where Alaska lags behind. When software makes decisions about loan approvals, job interviews, or benefits applications, fairness and transparency are crucial. Yet, unlike many other states, Alaska has no requirements for explainability, audibility, or disclosure in these processes. This absence of regulation leaves individuals vulnerable to discriminatory outcomes.

Disinformation is also a pressing concern in the age of AI-generated deepfakes. With roughly 30 states having passed election-related deepfake disclosure laws ahead of this year’s midterms, Alaska stands out as an exception. The state has yet to establish meaningful safeguards against AI-generated nonconsensual intimate imagery or fabricated videos and images.

A previous attempt to address these issues was House Concurrent Resolution 3, which aimed to create a task force studying AI in Alaska and recommending policies for its use in government, education, and everyday life. Although this effort stalled, the need for comprehensive governance has only grown more pressing.

Rather than abandoning HCR 3’s goals, proponents are advocating for a new approach: establishing research capacity at the University of Alaska Fairbanks. This proposal would create a graduate-level program incorporating computer science, public policy, philosophy, and ethics to study AI and data centers in an integrated manner. By examining these two areas together, researchers can better understand their interdependencies and develop more effective policies.

The relationship between AI and data centers is crucial for policymakers. As AI use grows, so does the demand for data centers – which, in turn, fuels further AI development. This feedback loop highlights the need to address both issues simultaneously. Any Alaska framework that takes AI seriously must consider its infrastructure requirements, including electricity draw on grids, water usage, and community effects.

The ultra-rich tech giants are largely driving AI policy without input from state governments or local communities. If Alaskans want a say in this process, they should contact their legislators to revive HCR 3’s task force with real teeth and support the University of Alaska Fairbanks proposal for research capacity. By doing so, Alaska can avoid being the last state at the table when it comes to AI governance.