A look at Anthropic's hiring process, which prohibits AI use in interviews and features a culture interview that candidates describe as highly intense (Jo Constantz/Bloomberg)
THE SO WHAT
Banning AI use in interviews and running intense culture screens is Anthropic saying 'we want humans who can think unaided about systems that might matter morally.' If you're hiring for core AI roles and letting candidates lean on tools during evaluation, assume you're selecting for tool fluency over original judgment.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIAnthropic details its progress toward recursive self-improvement, and its implications, and says 80%+ of the code merged into its codebase is authored by Claude (Anthropic)
When 80% of merged code is authored by Claude, the constraint in AI companies shifts from engineering capacity to oversight and product judgment. If recursive self-improvement is on the table, your real moat becomes evals, guardrails, and who controls the spec for what the system is allowed to optimize for.
Applied AI'At least they're honest about it?' —Tomb Raider: Legacy of Atlantis is the latest game to come with an AI-generated content disclosure
AI content disclosures are becoming part of the UX—players are being trained to expect and judge AI involvement. Studios that treat disclosure as a design surface, not a legal footnote, will control the narrative on where they use AI and why it improves the game instead of cheapens it.
Applied AICanada unveils its AI strategy that it says will help create 250,000 jobs by 2031, with a CA$500M tech fund to invest in homegrown AI startups (Nivedita Balu/Reuters)
Canada putting CA$500M behind AI with a 250,000-job target is industrial policy, not PR—talent and startups will follow the subsidy. If you’re building in or selling into Canada, assume procurement, grants, and immigration will increasingly favor “AI strategy compliant” businesses over neutral ones.
Applied AIHackers could use poisoned WhatsApp and Slack notifications to take over your Google Gemini – and make it work on their behalf
Prompt injection via Android notifications shows the attack surface is now every UX surface where text touches an assistant—not just the chat window. If you’re shipping agentic workflows, you need a red-team pass on notification channels and third-party integrations this week, or you’re delegating control to whoever can send a message.