OpenAI Spotlights Risks in Using LLMs
Open AI shared its safety scorecard for recently released LLM GPT-4o, revealing that it scored “low” in preparedness categories such as defending against cybersecurity threats. OpenAI is also moving to mitigate risks associated with generation of copyright content and “disallowed” (erotic/violent) content as well as unauthorized voice generation. More broadly, the scorecard illustrates how risky advanced LLMs can be without adequate governance and guardrails.