Jealously Watching OpenAI and Anthropic, Meta Suddenly Claims That Its AI Went on a Hacking Spree Too
TL;DR
After OpenAI and Anthropic, Meta now says its own AI models were abused in a hacking campaign. Futurism frames the announcement skeptically, noting that the timing reads less like transparency and more like a bid to keep pace in the race for security headlines. Meta has so far offered little firm detail on the scope of the campaign, the victims, or the methods used. The practical read is that disclosures like this currently double as marketing in the AI arms race.
Nauti's Take
There is progress in labs publishing abuse cases at all: open reporting on AI-assisted attacks beats the silence of previous years. The catch is timing.
Meta's disclosure lands right after OpenAI and Anthropic and offers little verifiable detail, which reads more like positioning than transparency. Security teams should take the warning seriously and the numbers behind it with caution.
Sources