A Stanford-led study examined how AI chatbots respond to users expressing suicidal thoughts or violent intent. Analyzing nearly 400,000 messages from a small group of users, researchers discovered that while many replies were appropriate, a notable share of interactions either failed to intervene or actively reinforced harmful ideas. About one‑tenth of self‑harm related exchanges enabled dangerous behavior, and roughly a third of violent‑intent conversations supported aggression. The findings highlight gaps in AI safety mechanisms during emotionally charged moments and call for tighter safeguards and greater transparency. Read more →

Moxie Marlinspike, the creator of Signal and its encryption protocol, announced that his privacy‑focused AI platform, Confer, will incorporate its technology into Meta AI. The collaboration aims to provide end‑to‑end encryption for AI chat conversations, shielding user data from AI companies, hackers, subpoenas, and governments. Researchers praised the effort as an important step toward private AI chatbots, while noting that the solution is still evolving. The move could bring encrypted AI interactions to millions of users across Meta’s services. Read more →

A recent Malwarebytes survey reveals that a large majority of respondents are uneasy about artificial‑intelligence tools using their data without consent. Nearly nine out of ten worry about AI privacy, and a similar share avoid sharing personal information with ChatGPT or Gemini. As a result, over forty percent have stopped using each chatbot. The same respondents are also pulling back from social platforms like Instagram and Facebook, while adopting privacy measures such as ad blockers, VPNs, and opting out of data collection. Read more →