AI Sec Watch

Anthropic announced Claude Science, a new AI product designed to help scientific researchers conduct work autonomously in areas like computational biology and drug development, similar to how Claude Code assists software engineers. The US also lifted restrictions on Anthropic's Mythos and Fable models after security discussions, though the delay has already benefited Chinese AI competitors.

Researchers discovered that SemAder, a technique that manipulates both the structure and meaning of binary code (compiled programs), can fool LLM-based binary code analysis tools into missing security problems. The study shows that by carefully modifying how code is organized and what it semantically does, attackers can evade detection systems that use large language models to analyze compiled programs for vulnerabilities.

Google released a new smart speaker designed specifically for Gemini (Google's AI assistant), marking the company's first new smart speaker in six years. While the hardware is well-designed, the article indicates that Gemini for Home, the AI software powering the speaker, still feels incomplete and not fully ready for users.

Abu Dhabi's MGX, a major investment fund, has closed a $49 billion fund to back AI companies, making it one of the largest investment vehicles in the sector. The fund has invested in major AI companies like OpenAI and Anthropic, and is looking to invest across the AI tech stack (the layers of technology needed to build AI systems, including hardware, software, and platforms). This reflects the massive amount of money flowing into AI companies, which have raised a record $416.6 billion so far this year.

Check Point Research discovered a Python Flask web application where an AI model independently combined a browser vulnerability with a working ransomware technique to create malware that operates entirely within a web page, requiring no exploit kit, software installation, or technical skill from the attacker. This represents a shift in AI-assisted threats, as the AI connected theoretical security risks to practical attack methods on its own.

Security researcher Ian Carroll used Claude Opus (an AI assistant) to discover a vulnerability in Front Gate Tickets' website that allowed him to gain super-administrator access (high-level control over a system) and issue free tickets to music festivals across the US. Carroll reported the bug responsibly rather than exploiting it, and Front Gate patched the vulnerability within 24 hours, demonstrating how AI tools can help identify serious security flaws in web systems.

Large language models frequently invent web addresses that don't exist, and attackers are now registering these fake domains before anyone else can, then hosting phishing pages on them to catch people following AI-generated links. Palo Alto Networks' research found that when two AI models answered 685,339 questions about major brands, they generated 2.1 million links, including roughly 250,000 made-up domains with no owner yet and 13,229 known-malicious addresses. This attack, called phantom squatting (a technique where criminals register fake domains that AI systems invented), works because brand-new domains have no reputation history for security filters to flag them, so victims reach the phishing site before defenses catch up.

Anthropic restored access to Claude Fable 5 worldwide after the U.S. lifted export controls (restrictions on who can use a technology) that had been imposed for two and a half weeks following the discovery of a jailbreak (a prompt that tricks an AI into bypassing its safety rules). To address the security concern, Anthropic trained a new safety filter called a classifier that blocks the specific jailbreak technique in over 99% of attempts, while automatically routing blocked requests to a weaker model and notifying users.

Anthropic's Claude Fable 5 and Mythos 5 AI models, which the U.S. government had restricted from being shared with foreign nationals due to national security concerns, have been cleared for release again following negotiations with the Trump administration. Fable 5 will be available globally starting Wednesday through Claude's platforms, while Mythos 5 access has been restored for some U.S. organizations with plans to expand through Anthropic's Glasswing program (a cybersecurity initiative providing selected organizations access to advanced AI models for defensive security testing).