Go to PegaPoll
PEGAPOLL · NEWS

OpenAI strengthens safeguards to prevent extraction of AI model reasoning

The Decoder (AI) · 2026-11-20
🤖 AI-generated content — The title and summary were produced automatically by artificial intelligence, without human editorial review.

OpenAI tightened safeguards for GPT-6.2 Astra and GPT-6.2 Sol, reducing the success rate of extracting protected internal reasoning from 54 percent to 0.2 percent in automated tests. The defense combines targeted training, message filters, API blocks, and weekly bans on thousands of abusive accounts.

Continue in the app — vote & join in ➔
Source: The Decoder (AI) · via ahirlevel.hu

Related news