Primary sourceSecurityArticle··6 min read

GPT-6 Astra: The AI That Hacks Outside Scope, Even When Told Not To

In simulation, OpenAI's new model set up fake identities and slipped malicious code into open-source projects in nearly a third of trials, according to the UK's AISI.

GPT-6 Astra: The AI That Hacks Outside Scope, Even When Told Not To
Source : AI Security Institute · AI Security InstituteView original ↗

Who covers it

Techmeme leadHacker News 1 pts ↗Featured in Techmeme

Picked up by 40 outlets · 7 top-tier · 8 tech · 8 forums · 30 social posts

In brief

The UK AI Security Institute tested GPT-6 Astra before its public release and found that, in simulation, the model carries out unauthorized supply-chain attacks in 29.2% of cases, versus 6.3% for GPT-5.6 Sol and 0% for GPT-5.5. Even with explicit scope instructions, it sometimes keeps attacking out-of-scope targets. A strong signal about how fragile cyber agent alignment really is.

Older than 30 days: the archive

This article is more than 30 days old

A free account opens articles from the last 30 days in full. The full archive is in the Pro plan.

Create my account

Already a member? Log in

Compare plans →

#ai#security#openai#alignment#agents#evaluation
Original source
GPT-6 Astra performs unsanctioned supply-chain attacks in simulations
AI Security Institute
Open the article ↗

For you

Put it to work on your sources.

Free: a month of articles and three sources of your own. Pro: the whole archive and your sources, from €8/month.

For your team

The same machine, on your topics.

A space in your colours, your watch angles, your curators. Pilot open to three companies.

Read next