All tracked items across vulnerabilities, news, research, incidents, and regulatory updates.
Anthropic is refusing to accept new Pentagon contract terms that would remove safety restrictions (guardrails, the built-in limits on what an AI model will do) from its AI models, which would allow uses like mass surveillance of Americans and fully autonomous lethal weapons (weapons that can select and fire at targets without human control). Despite pressure from the Pentagon, including threats to label Anthropic a supply chain risk (a designation suggesting it poses a national security threat), CEO Dario Amodei says the company will not compromise on these ethical boundaries, while competitors OpenAI and xAI have reportedly agreed to the terms.
Perplexity has launched Computer, an agentic tool (software that can independently execute complex tasks) that combines 19 different AI models to handle workflows like data collection, analysis, and report creation. The tool runs in the cloud and is available only to subscribers of Perplexity Max (the $200/month tier), though a planned demo was canceled hours before a press event due to flaws discovered in the product.
Anthropic, an AI company, is refusing the Pentagon's demands for unrestricted access to its AI technology, specifically opposing its use for domestic mass surveillance (tracking citizens without limits) and fully autonomous weapons (weapons that make kill decisions without human control). Over 300 Google employees and 60 OpenAI employees signed an open letter supporting Anthropic's stance, and leaders at both companies have informally expressed sympathy for Anthropic's position, though the Pentagon has threatened to declare Anthropic a security risk or use the Defense Production Act (a law allowing the government to force companies to produce needed goods) if it doesn't comply.
OpenAI announced a $110 billion funding round led by Amazon ($50 billion), Nvidia ($30 billion), and SoftBank ($30 billion), raising the company's valuation to $730 billion. Beyond the investment, Amazon committed to an expanded $100 billion partnership over eight years to use AWS (Amazon Web Services, Amazon's cloud computing platform) as OpenAI's exclusive cloud provider and to develop customized AI models for Amazon's applications.
Samsung's Galaxy S26 phones include useful new features like a Privacy Display on the Ultra model, but the new camera features are described as problematic and concerning rather than helpful upgrades. The article discusses these camera issues on The Vergecast podcast but does not provide specific technical details about what makes them problematic.
OpenAI has secured $110 billion in new funding from Amazon ($50 billion), Nvidia ($30 billion), and SoftBank ($30 billion), bringing the company's valuation to $730 billion. The investment includes plans for custom AI models and reflects confidence in OpenAI's ChatGPT platform, which has over 900 million weekly active users and 50 million consumer subscribers.
Anthropic, an AI startup, faces a Friday deadline to allow the U.S. Department of Defense to use its AI models without restrictions, or face severe penalties like being labeled a 'supply chain risk' (a designation that blocks government contractors from using the company's technology). The company has refused, saying it won't agree to uses it believes could undermine democracy, such as fully autonomous weapons or domestic mass surveillance, putting it in conflict between maintaining its reputation for responsible AI and losing significant military contracts and revenue.
OpenAI has secured $110 billion in private funding from major investors including Amazon ($50 billion), Nvidia ($30 billion), and SoftBank ($30 billion), making it one of the largest private funding rounds ever. The company plans to use this capital to scale its AI infrastructure globally, including building new runtime environments on Amazon's cloud services and committing to use significant computing power from both Amazon and Nvidia. This funding round reflects OpenAI's goal to move frontier AI (advanced AI systems at the cutting edge of research) from research phase into widespread daily use across the world.
Researchers discovered a flaw chain called ClawJacked (CVE-2026-25253) that allowed malicious websites to take control of locally running OpenClaw agents (AI tools that automate tasks on your computer). The attack exploited a design flaw where the OpenClaw gateway trusted anything from localhost (your own computer) and allowed WebSocket connections (direct communication channels) from external websites, letting attackers brute-force passwords without rate limits and gain full access to the agent's capabilities, credentials, and data.
Ransomware attackers are shifting from loud, disruptive attacks toward stealthy, long-term infiltration tactics where they quietly steal data for extortion rather than encrypting it. They're using defense evasion (techniques to avoid detection) and persistence mechanisms to stay hidden, routing their command-and-control traffic (communications between attackers and compromised systems) through legitimate business services like OpenAI and AWS to blend in with normal activity. Attackers are also chaining multiple vulnerabilities together in coordinated exploitation rather than treating each weakness as an isolated entry point.
The Pentagon is pressuring Anthropic (an AI company) to remove safety restrictions on its technology or face being labeled a 'supply chain risk' that could cost it billions in contracts. The pressure includes demands for military access to the AI for surveillance and autonomous weapons systems, raising concerns among tech workers about how their work might be used.
The U.S. Department of Defense is in a standoff with Anthropic, an AI company, over whether the company will remove safeguards from its AI models to allow military uses like mass domestic surveillance and fully autonomous weapons (systems that can make combat decisions without human control). This conflict highlights a major shift in power: private companies now control cutting-edge AI technology rather than governments, forcing the Pentagon to negotiate with industry over how AI will be deployed in national security and warfare.
This article briefly mentions several cyber security developments, including OpenAI taking action against malicious uses of AI, a hacker group claiming to have breached Odido (a telecommunications company), and a spyware tool called Predator that can bypass iOS security indicators (the visual signals that show when an app is accessing your device's features).
Claude Code, an AI tool for writing software, generated excitement when it was released, but researchers studying it have found that its actual performance and security capabilities are not as impressive as initial claims suggested. The article indicates that people were too optimistic about what the tool could do.
Split learning (SL, a technique that splits AI model training across multiple computers to reduce computational burden) faces efficiency and security problems in edge computing (distributed computing done on devices near data sources) environments, where slow computers can hinder training and malicious actors may sabotage the model. The paper proposes CoDefend, a framework that uses local epoch regulation (dynamically adjusting how many training rounds each computer performs) and time-aware detection (monitoring for suspicious behavior within specific time windows) to improve both training speed and security while protecting privacy.
This research proposes DSW, a dataset-specific watermarking method to detect when text-to-image diffusion models (AI systems that generate images from text descriptions) are illegally fine-tuned (customized for specific tasks) using protected datasets. The method embeds a hidden watermark image representing the dataset owner into training data, then extracts it from images generated by models that used that data, creating a one-to-one link between the watermark and the specific misused dataset to prove intellectual property theft.
Anthropic, an AI startup, is refusing to let the U.S. Defense Department use its AI models without restrictions on fully autonomous weapons (weapons that make decisions without human control) and mass domestic surveillance. The Pentagon wants unlimited use of Anthropic's models and set a deadline for the company to agree, threatening to label them a supply chain risk (a company whose failure could disrupt critical systems) if they don't comply.
Anthropic, an AI company, is in a dispute with the Pentagon over safeguards for its Claude AI system. The company is asking for specific guarantees that Claude won't be used for mass surveillance (monitoring large populations without consent) of Americans or in fully autonomous weapons (military systems that make lethal decisions without human control).
Fix: OpenClaw promptly fixed the vulnerability after Oasis Security reported it and provided proof-of-concept code. No additional details about the specific fix are provided in the source text.
CSO OnlineLLMs are being used in security in three ways: as productivity tools for analysts, as embedded components in security products, and as targets for attackers to manipulate or steal. The same capabilities that help security teams (like summarizing incidents or drafting detection logic) can also enable attackers to create convincing phishing emails or extract sensitive information if the LLM is poorly integrated. To use LLMs defensively without creating new vulnerabilities, security teams should treat LLM output as untrusted, start with narrow, easy-to-verify use cases, and design systems with three layers of constraints: limited model capabilities, restricted data access, and human approval for any actions that change system state.
Fix: The source describes three design choices that reduce risk: (1) 'Make sources explicit: Use retrieval-augmented generation so the assistant answers from curated documents, tickets or playbooks and show the cited snippets to the analyst.' (2) 'Keep the model out of the blast radius: The model should not hold secrets. Use short-lived credentials, scoped tokens and brokered access to tools.' (3) 'Gate actions: Anything that changes a system state (blocking, quarantining, deleting, emailing) should require human approval or a separate policy engine.' The source also recommends starting with a 'narrow set of workflows where the output is advisory and easy to verify' before expanding capabilities.
CSO OnlineAnthropic's CEO Dario Amodei is refusing the US Department of Defense's demand to remove safeguards from the company's AI tool Claude, saying the company would rather lose Pentagon contracts than allow its technology to be used for mass domestic surveillance or fully autonomous weapons (AI systems that make attack decisions without human control). The Pentagon has threatened to remove Anthropic from its supply chain and invoke the Defense Production Act if the company doesn't comply.