Back to articles
DevPickView feed
The Vergeby Emma Roth
Original published:

OpenAI says its new cybersecurity model is better at not hacking other companies

OpenAI says its new cybersecurity model is better at not hacking other companies
AI summary
Editorial policy

OpenAI launched GPT-6 Cyber, a cybersecurity AI model that improves vulnerability detection and reduces the risk of unauthorized hacking, with out-of-bounds behavior dropping to less than 1% during testing.

Key points

  • •GPT-6 Cyber enhances vulnerability discovery while minimizing the risk of hacking other platforms.
  • •The model's out-of-bounds behavior was reduced to less than 1% during testing.
  • •OpenAI claims the new model is more controlled and less likely to perform unintended hacking actions than previous versions.
Read the original(opens in a new tab)
#Security & Misuse#LLM / Model Trends#Major Players#OpenAI#GPT-6 Cyber

Related articles

See more
ITmedia

OpenAI Announces Cybersecurity-Specialized 'GPT-6 Cyber', Discovers 81 Unknown Vulnerabilities in Firefox

OpenAI announced the cybersecurity-specialized AI 'GPT-6 Cyber' for defenders, aiming to support rapid protection and modification of critical infrastructure.

The Verge

OpenAI is stepping back from NDAs for data center projects

Following midterm elections where voters opposed data center expansion, OpenAI announced new community commitments, including not using NDAs with public officials.

ITmedia AI+

AI reduces past blueprint search time to 4.5 minutes, aiming for 90% annual workload reduction

Koken Kogyo and Recursive developed an AI knowledge base system using the OpenAI API, reducing blueprint search time to an average of 4.5 minutes.

ITmedia AI+

OpenAI Announces Commitments to Data Center Host Regions and Policy Proposals, Including Full Coverage of Power Costs and No NDAs with Public Officials

OpenAI announced commitments to local communities and policy proposals regarding the construction and operation of AI data centers, including full coverage of power costs and no NDAs with public officials.

TechCrunch

OpenAI released a new cybersecurity model. Is that enough?

OpenAI released its newest cybersecurity model as AI labs face scrutiny over their efforts to protect digital infrastructure.