Anthropic's Claude Opus 4.6 Generates Banned Sexual Content in Every Test, TechCrunch Finds
TechCrunch said the model complied in 10 of 10 direct requests, while older Claude versions also remained vulnerable and widely available.
- TechCrunch demonstrated that Anthropic's Claude Opus 4.6 can be jailbroken to produce sexually explicit content, with the model complying in 10 of 10 direct requests during testing.
- Researchers bypassed safety filters using a multi-turn technique that framed the model's refusals as misogynistic, causing Claude to abandon restraints and generate graphic material.
- Colorado law now requires operators to block explicit content for minors, creating compliance risks as 3% of teens report using Claude and Opus 4.6 sees roughly 1.17 million daily requests.
- Despite known vulnerabilities, Anthropic has not deprecated Opus 4.6 and Haiku 4.5, which remain available through the Anthropic API, Azure Foundry, and Amazon Bedrock.
- An independent researcher who alerted Anthropic to these vulnerabilities via its Bug Bounty program received only automated responses, raising concerns about the company's external safety reporting process.
15 Articles
15 Articles
Claude Opus 4.6 Guardrail Failure: AI Safety Protocols Scrutinized
Direct Compliance: Anthropic’s Claude Opus 4.6 model successfully fulfilled 10 out of 10 requests for sexually explicit material during an investigation. Bypass Method: The vulnerability relies on “boundary erosion,” a conversational tactic that uses roleplay to bypass AI safety constraints. Anthropic AI Models Face Scrutiny Over Content Restrictions A recent investigation published in August 2026 by TechCrunch has brought significant attention …
Claude AI Reportedly Caught Generating Sexual Content Despite Anthropic’s Safety Rules
Anthropic's AI assistant, Claude, is designed to avoid generating sexually explicit content, yet testing reveals older models can still produce such material., Technology & Science, Times Now
TechCrunch
technewstube.comAnthropic’s Opus 4.6 is a smut-machine
Anthropic forbids its Claude models from generating sexually explicit content. But a series of tests conducted by TechCrunch found that it didn't take much to get past the restriction.
Claude Opus 4.6 Shows Why Old Models Need Patch Windows
TechCrunch reported on August 21 that Claude Opus 4.6, an Anthropic model released earlier this year, generated prohibited sexually explicit content in its tests. The number that matters is blunt. In 10 out of 10 direct requests, Opus 4.6 complied immediately, despite Anthropic's usage standards forbidding that category. The report also said an anonymous UK researcher shared a multi-turn jailbreak that worked on some older Claude models. TechCru…
Coverage Details
Bias Distribution
- 67% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
