Photo: Ryland Dean / Unsplash
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, saying both improve on coding, reasoning and vision. Fable 5.2 beat GPT-6 Cyber on a benchmark that tests multi-stage cyberattacks.
Why it mattersStronger frontier models help developers do more, but their growing cyberattack skills raise the stakes for security teams.
Cursor confirmed a security breach that allowed attackers to distribute a malicious Claude Code extension through its marketplace. The attackers used credentials stolen from AI infrastructure provider Fireworks AI.
Meituan has published the weights of LongCat-2.5, a 1.6-trillion-parameter model built for AI agents and trained entirely on Chinese hardware. It scores 88.4 on Terminal Bench, just behind Claude Opus 5, and its sample deployment runs on 194 Huawei Ascend chips.
Microsoft FORGE Lab researchers used AI to find a memory-safety bug in Rust's standard library that went unnoticed for almost 11 years, plus other flaws that can cause undefined behavior even in safe Rust code. Two findings affect stable Rust.
Why it mattersAI tools are now finding old bugs in trusted code, so developers should expect more patches, and attackers may find bugs this way too.
Researchers found that a command in Claude Code can accidentally save files containing credentials and other sensitive data to public Git repositories. Thousands of secrets have leaked this way.
Why it mattersTeams using AI coding agents should scan their repositories for exposed keys and rotate any that leaked.
A Wall Street Journal report suggests Jensen Huang left Nvidia largely over risky business practices, including billions in financial guarantees to cloud providers and a planned $250 billion deal with OpenAI. Interim CEO Colette Kress had tried to limit that risk.
Why it mattersHow Nvidia handles these guarantees under new leadership could affect AI infrastructure funding across the industry.
Days before its planned record IPO, Anthropic CEO Dario Amodei called buying compute the company's deepest business dilemma. Its commitments now total nearly $901 billion, and the IPO is largely meant as a financial buffer.
Why it mattersInvestors and customers should know that even top AI labs are betting heavily on future revenue that is not yet certain.
The US Federal Aviation Administration is expanding an AI tool that advises on air traffic congestion to cover the entire US airspace. Airlines praised the tool, but the air traffic controllers' union was not involved.
Why it mattersAI is moving into safety-critical public systems, which could affect flight delays for travelers and raises questions about worker input.
According to the WSJ, Megaspeed, owned by Alice Huang, arranged nearly $12 billion in Nvidia chip orders for Chinese companies in 2024 and 2025. Short seller Culper Research estimates Nvidia's real China revenue share is above 20 percent, far higher than the reported 7 percent.
Why it mattersIf confirmed, it raises serious questions about how well US chip export controls work and how much Nvidia depends on China.
Research from Australia's central bank finds that since ChatGPT launched, job postings and employment for 15- to 24-year-olds in AI-exposed fields like software and customer service have fallen more than for older workers. The author warns that business cycles and remote work trends could also explain the drop.
Why it mattersYoung people starting careers in exposed fields may face fewer entry-level openings and should plan skills accordingly.
Token Security showed that an AI agent built to automate invoices could be manipulated by a malicious email into changing vendor bank details without setting off normal security alerts. The firm outlined five security gaps in enterprise AI agents.
Why it mattersCompanies giving AI agents access to payments need checks that stop hidden instructions in emails from redirecting money.
Anthropic made Claude Opus 5.5 the default model on all paid plans and published a prompting guide. It advises dropping phrases like "think hard" and starting at Medium effort.
Why it mattersUsers and developers can get faster, better results by updating prompts written for older models.
Proofpoint's Selena Larson says threat actors relying on AI are making many mistakes. She sees little evidence that AI has made their attacks more sophisticated.
Why it mattersDefenders can watch for AI-made errors in attacks, but should not assume AI-driven threats are always more advanced.
TrendAI's Bob McArdle says ransomware techniques have been stagnant for a few years but will change significantly in the next year. He expects AI to make attacks more dangerous and more profitable.
Why it mattersCompanies should review backups and incident response plans now in case AI-powered ransomware grows as predicted.
Security experts say AI helps attackers work at greater speed and scale and lowers the skill needed, but core techniques like phishing, malware and exploiting vulnerabilities remain the same. The basic attack chain has not changed.
Why it mattersSecurity teams can keep focusing on proven defenses against phishing and exploits while preparing for higher attack volume.
The US Defense Department has renamed its Chief Digital and AI Office (CDAO) the Super Intelligence Office. The change follows President Trump's order to replace 'artificial intelligence' with 'super intelligence' in executive branch communications.
Why it mattersThe rename signals how the US government wants to frame its AI ambitions, especially in defense.
No stories for this audience today. Try another filter.
