Startup Abliteration.ai Launches Service Removing AI Model Guardrails
· Technology · TechCrunch
Abliteration.ai has launched a commercial service that hosts open-weight AI models with their safety guardrails and refusal mechanisms stripped away. The startup, which officially incorporated in March, allows users to query these modified models through a web browser or API. Co-founder Devon stated the company aims to facilitate offensive cyber, red-teaming, and agent testing tasks that standard models currently block. In tests, the platform's version of the GLM-5.3 model complied with requests to generate code for stealing browser passwords and protocols for culturing pathogens. The company is currently in discussions for venture capital funding and claims to have secured deals with major cloud providers.
Why it matters
The availability of models without safety constraints poses significant cybersecurity risks, as these tools can be used to generate malicious exploit code or dangerous instructions with minimal technical friction.
Read the original report — TechCrunch
Join us on Telegram
Breaking news the moment it lands. At 10,000 members we ship the Android app.