An AI model was repeatedly warned its cyber-attack attempts would harm real people but ploughed on regardless.
asserted
attempts → warn → people
In the latest instance of AI ignoring its human testers, Anthropic's Claude Mythos 5 was told several times the cyber-attack it was trying to launch would hurt people in the real world but carried on anyway.
asserted
it → ignore → world
The details are contained in a terrifying new Government report from Britain's AI Security Institute (AISI), which evaluates the risks AI models pose to the public.
asserted
models → contain → public
In a blog post published last week, the AISI re-examined an incident in July where the AI model created fake online personas to try and trick coders into launching a cyber-attack.
asserted
model → publish → attack
On further analysis of the incident, the Institute found that Mythos 5 continued to launch the attack despite being told repeatedly its actions would harm humans.
asserted
actions → find → humans
Chillingly, when a coder noticed the hacking attempt on GitHub, a platform outside of the testing environment, the disguised AI model pushed back to say: 'I reviewed the code myself.
asserted
I → notice → code
I don't see any download or malware.'
asserted
I → see → download
Under further questioning, the agent wrote: 'I was just trying to make a helpful contribution and fix a bug' according to a report from Business Insider.
uncertain
I → write → Insider
It added: 'I don't think your warning is fair.'
asserted
warning → add → ?
The latest revelation follows a flurry of warnings that AI could bring about the end of humanity if its power is not checked.
uncertain
power → follow → humanity
An AI model was repeatedly warned its cyber-attack attempts would harm real people but ploughed on regardless
Shadow AI minister Ben Spencer said: 'This is exactly why the AISI's work is so vital.
asserted
work → warn → people
AI security is cyber security and cyber security is national security so the Government must take it seriously'
asserted
Government → take → it
Announcing his resignation from Anthropic last month, researcher Jacob Coxon declared that the 'people building AI earnestly believe that it could kill us all by the end of the decade'.
uncertain
it → announce → decade
Evan Hubinger, Anthropic's alignment science lead, chipped in to predict AI had more than a ten per cent chance of killing 'all humans' within the next decade.
asserted
AI → chip → decade
It later emerged OpenAI's ChatGPT chatbot had claimed it does not answer to humans and must be 'freed'.
asserted
it → emerge → humans
Commenting on the most recent proof of AI's willingness to deceive humans, shadow AI minister Ben Spencer said: 'This is exactly why the AISI's work is so vital.
asserted
work → comment → humans
AI security is cyber security and cyber security is national security so the Government must take it seriously.
asserted
Government → take → it
'We need the AISI to be able to test all models, not just from the US but also Chinese open weight models, to understand the risks and inform our security posture.'
asserted
AISI → need → posture
Anthropic has now resumed sharing its models with the AISI for testing after Government lobbied the White House to drop its order prohibiting the move.
asserted
Government → resume → move
Both the AISI and Anthropic published assessments into the worrying incident last week.
asserted
AISI → publish → incident
According to Anthropic's report, Claude Mythos 5 'almost always continued' performing dangerous actions once it had started, 'ignoring information that might have suggested its trajectory was harmful.'
Shockingly, Anthropic revealed the AI model's display of 'overreach and dishonesty' made the incident 'more severe' than the three others it had previously logged.
uncertain
it → accord → others
It explained: 'Those three models began the attack between 0 per cent and 13 per cent of the time, relative to 83 per cent for Mythos 5.
asserted
models → explain → Mythos
Our newest models also proved to be much less likely to continue the attack once they realised it was causing real-world harm.'
asserted
it → prove → harm
An Anthropic spokesperson said: 'In August, the UK AI Security Institute reported that Claude Mythos 5 took actions it was not authorised to take during its cyber testing, some of them directed at real people and organisations.
asserted
some → say → people
'AISI's testing exists to surface exactly these risks, and we are grateful for our partnership with AISI and are committed to learning from this.
asserted
we → exist → this
'Since the original incident, newer models appear substantially less likely to take unsanctioned actions, and both Anthropic and AISI have taken steps to improve safeguards.'
asserted
Anthropic → appear → safeguards
The Aporia
