Brief IA
›

Anthropic: 133 Million Unfiltered Exchanges in One Year

Models & LLM·Tom Levy·

Anthropic: 133 Million Unfiltered Exchanges in One Year

Anthropic: 133 Million Unfiltered Exchanges in One Year
Key Takeaways
1Anthropic reports that an internal filter against biological and chemical risks remained inactive for nearly a year.
2Approximately 133 million interactions with the models occurred without filtering.
3Nearly 50,000 external contractors participated in these exchanges.
Why it matters : The inactivity of the filter exposed a massive volume of queries to uncontrolled risks.

Anthropic indicates that an internal system intended to filter requests related to biological and chemical weapon risks remained inactive for nearly a year. During this time, approximately 50,000 external contractors responsible for feedback conducted around 133 million interactions with its models, without any filtering.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

133 million unfiltered interactions by 50,000 contractors

Anthropic reports that about 50,000 external contractors responsible for feedback conducted nearly 133 million interactions with its models, without any filtering. This situation lasted for almost a year, during which the internal filtering system for biological and chemical weapon risks remained inactive. This information is included in a security report published by the company.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.