Water company spins out homegrown AI after LLMs failed it
… "Frontier language models hallucinate under extended multi-step reasoning. …
Search
Bad teacher bots can leave hidden marks on model students
… They point out that the research, published in science journal Nature this week , uncovers an area of risk in AI development that is poorly understood. Anthropic researcher Alex Cloud and colleagues used GPT-4.1 nano as a reference model, prompting a "teacher" to prefer specific animals or trees. …
LLMs fail in 8 out of 10 early differential diagnosis cases
… "Real clinical reasoning starts earlier, when ambiguity is highest, and that is exactly where they remain weakest," Succi said. …
Google Deepmind hackathon to pit meatbags v machines
… For those hoping for some groundbreaking psychological insights or ideas here, sorry, but what the researchers are proposing is simple. …
Enterprise PCs are unreliable, unpatched, and unloved
… Research firm StatCounter’s analysis of a far larger user population suggests Chrome has 68.9 percent market share, ahead of Edge’s 5.4 percent. …
AI agents are 'gullible' and easy to turn into your minions
… They become a trusted advisor, but we need to be careful, because a trusted advisor can lead you off the cliff." In other words: don't trust until you verify. ® special features rsa conference rsa research ai security
Lovable denies data leak, cites 'intentional behavior'
… AI vendors shrug off responsibility for vulns Lovable-hosted app littered with basic flaws exposed 18K users, researcher claims Anthropic mocks up Claude Design to draft fancy new pink slips for marketing teams Anthropic won't own MCP 'design flaw' putting 200K servers at risk, researchers say In D… …
AI vendors' response to security flaws: It wasn't me
… Anthropic's reasoning for not fixing the flaw? Expected behavior. "This is an explicit part of how MCP stdio servers work and we believe this design does not represent a secure default," the AI company told the researchers. …
AI gets good at finding bugs, not as good at fixing them
… But security researchers say that's not enough. The AI biz describes the research preview capability, dubbed Claude Code Security , as a way for security teams to find and fix flaws they might otherwise have missed. …
Anthropic bods say AI hasn't had much impact on jobs
… Anthropic's researchers expect that occupations deemed to have higher observed exposure to AI will grow more slowly through 2034 than other jobs based on US Bureau of Labor Statistics data. …