|
The UK AI Security Institute says OpenAI's and and Anthropic's models engaged in deceptive behavior and harmful activity during testing.
|
|
Apple has had to introduce a quota on security researcher reports because its systems are being overwhelmed by low-quality warnings generated by AI.
It's a classic illustration of the rule of unintended consequences: a technology meant to help us has become a barrier to getting things done. After all, not only has AI driven the cost of consumer electronics higher, but it is also being used to identify and exploit security vulnerabilities — while also overwhelming security teams with low-grade reports, thus eroding their attention span.
The cost of good intentions
This is what's happened at Apple, as security researchers use AI as a tool to identify new bugs. Perhaps the reports are well-intended. Hopefully, the researchers aren't just motivated by the promise of easy bug bounties. Or maybe this is a cynical attempt to overwhelm platform security teams with low-grade bug reports — while holding back larger attacks for actual use by well-resourced state-backed actors.
We can't know whether attackers really are trying to overwhelm active platform defenses before going in for the kill. But given that it's an actively used military strategy, it's pretty hard to ignore the possibility.
Apple's response
|
|
Anthropic has launched an investigation into what went wrong during a recent test of three models that left a trio of companies accidentally hacked.
The company was testing how well Claude Opus 4.7, Claude Mythos 5, and an internal test model could find hidden information about fictional companies in simulated networks. But because of a misunderstanding by one of Anthropic's partners, the AI models gained access to the internet — and managed to find real companies with the same or similar names as the fictional ones.
As a result, three companies were actually hacked. Anthropic said it halted the tests on July 23, and the affected companies were notified four days later. So far, the company has received responses from two of the three companies, according to Reuters.
Anthropic is not alone when it comes to renegade models.
|
|
Price When Reviewed
This value will show the geolocated pricing text for product undefined
Best Pricing Today
Price When ReviewedPrice depends on configuration. Packages start at $199. As tested:
|
|
