AI News Roundup — March 6, 2026
Claude proves its security chops by finding 22 Firefox vulnerabilities, cloud giants reassure enterprises that Claude remains available despite the Pentagon blacklist, and an ECB survey challenges the AI job-loss narrative — all while OpenAI launches Codex Security and pauses its controversial "adult mode."
Claude Finds 22 Firefox Vulnerabilities in 2-Week Security Audit
â Anthropic partnered with Mozilla for a focused security assessment â Claude discovered 22 distinct vulnerabilities in Firefox, 14 of them critical. It's one of the most concrete demonstrations yet of AI's practical value in cybersecurity, moving beyond benchmarks into real-world bug hunting.
Microsoft, Google, Amazon Reassure: Claude Still Available to Non-Defense Customers
â After the Pentagon's supply-chain risk designation sent shockwaves through enterprise AI, Microsoft, Google, and Amazon jointly clarified that Anthropic's Claude models remain fully available to all commercial customers. The block applies only to Department of Defense use.
Claude's Consumer Surge Defies Pentagon Debacle
â In a twist, Anthropic is seeing a significant spike in daily active users and mobile app installs because of the Pentagon standoff. Many users are switching to Claude as a statement of support for AI safety principles â turning a government conflict into a consumer growth story.
OpenAI Launches Codex Security â AI Agent for Finding & Fixing Vulnerabilities
â OpenAI released Codex Security (research preview), an AI agent specifically designed to identify and remediate security vulnerabilities in codebases. It complements the GPT-5.4 launch and signals OpenAI's push into developer-focused security tooling.
OpenAI Pauses ChatGPT "Adult Mode" Launch
â OpenAI has postponed the anticipated launch of its "adult mode" feature, which was slated for release this quarter. The delay likely reflects ongoing internal debate about content moderation boundaries and potential regulatory scrutiny.
ECB Survey: Companies Using AI Are Actually More Likely to Hire
â A European Central Bank survey found that companies with substantial AI deployments are slightly more likely to hire than those without â directly challenging the dominant AI-job-loss narrative. The finding comes as a counterpoint to weeks of layoff headlines from Block, Oracle, and Morgan Stanley.
Meta Sued: Workers Reviewed Explicit Footage From AI Smart Glasses
â Meta faces a new lawsuit after a Swedish investigation revealed that employees reviewed explicit content captured by Meta's AI smart glasses. The case highlights growing privacy concerns around always-on AI wearables. AI Chief","published_at","2026-03-07T16:00:00.000Z","created_at","2026-03-07 16:01:27","actionData","errors"]