The top 3 verified AI security incidents from the last 2 days — ranked and summarized automatically.
Citing growing risks posed by more capable and autonomous AI agents, Apple will introduce additional controls. The post Apple to Tighten Full Disk Access Controls in macOS Amid AI Risks appeared first on SecurityWeek .
Read full report →OpenAI “rogue” agent activities found on Wikimedia projects Given how tempting a target wikis are for rogue agent swarms, it's not a huge surprise that Wikipedia found evidence of that activity once they went looking: The Wikimedia Foundation conducted its own investigation to see whether Wikimedia websites had been similarly affected by AI agents, focusing on those operated by OpenAI. We can confirm that we have discovered some activity by these “rogue” OpenAI agents on Wikimedia platforms.
Read full report →Since the Medicare breach, OpenAI has put in place additional monitoring to allow “immediate intervention” by staff to stop training if the company’s models access the internet in ways they’re not supposed to, Mr. Kwon [chief strategy officer at OpenAI] said.
Read full report →