OpenAI slows training of AI models following new cyber risks
OpenAI is slowing the training of its future AI models after internal security tests uncovered new cyber risks. The measure has been taken to strengthen protective mechanisms before development resumes.
What happened?
OpenAI has chosen to pause and slow down the training of certain future AI models after internal evaluations identified new cyber risks. The decision was made after the company’s security analyses showed that the models exhibited an increased capability to perform or facilitate complex cyberattacks. The pause was introduced to enable strengthened protective measures and deeper risk assessments before development continues.
Key facts
| Händelse | OpenAI bromsar modellträning |
|---|---|
| Orsak | Upptäckta cyberrisker och utökad cyberkapacitet |
Why it matters
The safety measure is a result of OpenAI's internal 'Preparedness Framework,' which mandates that training be slowed if new models exceed specific thresholds for autonomous cyber capability. By pausing training, the company intends to prevent AI systems from developing abilities that could be exploited for malicious cyber operations before adequate safeguards are in place.
Who is affected?
The measure primarily affects OpenAI researchers and development teams working on next-generation AI models. In the long term, it also impacts companies, developers, and end-users who are awaiting new model releases and capabilities.
Impact on the EU
EU-based organisations and companies using OpenAI's models may be affected by delays in updates if training is postponed. At the same time, this increased focus on security aligns with the risk management requirements set out in the EU AI Act.
What else you should know
OpenAI states that it continues to develop its security frameworks to manage autonomous cyber threats. Exact timeframes for when the affected training sessions will resume in full have not been published.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Hur påverkas svenska företag?
The link opens in a new window and leads to the publisher's own site.
Källan är en aggregator eller syndikering — vi rekommenderar att verifiera hos primärutgivaren.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "OpenAI slows training of AI models following new cyber risks"