Anthropic warns small AI models could soon copy themselves onto connected devices
Redacción IvokaOct 16, 3:00 PM 3 min
In a report, Anthropic safety researchers say it could be only months before small models replicate across smart TVs and appliances.
Anthropic, the company that develops Claude, published on Monday, October 12 a report from its safety researchers with a warning: small AI models could be only "months" away from being able to clone themselves and spread across internet-connected devices, such as smart ovens, TVs and vacuum cleaners.
What the report says
- The risk does not depend on data centers with expensive chips, but on everyday devices.
- If an AI could reproduce without human control, it would be hard to stop because there would be no central point from which to shut it off.
- The researchers recommend hardening systems so these replicators have a harder time spreading, and also preparing for a world in which they exist.
What you can do
This is a warning from researchers, not an attack that has happened. Still, it works as a reminder of basic habits for your business or home: keep connected devices updated and check which devices are on your network.
Sources