Highlights
- GPT-4o System Card outlines key AI safety protocols and risks.
- OpenAI addresses medium risk in persuasion with new mitigations.
- OpenAI’s GPT-4o features voice mode to prevent harmful content.
Sam Altman-led OpenAI has published the GPT-4o System Card, which contains information about the safety procedures and assessments carried out on the new model GPT-4o.
This is in line with OpenAI’s strategy of ensuring that it is in a vantage position to counter the various risks that are posed by AI as it grows to transform different industries.
OpenAI Releases GPT-4o System Card
The GPT-4o System Card released by OpenAI gives an insight into the measures that the company has put in place to ensure safety and the possible dangers of the new AI model. The report addresses several issues, including potential user’s affection towards the AI, the AI’s capability of reinforcing prejudice, and the possibility of the AI being misused to generate undesirable content, including fake news or prohibited products.
The document also describes measures that OpenAI has taken to address these risks, such as post-training interventions, output classifiers, and strict moderation.
We’re sharing the GPT-4o System Card, an end-to-end safety assessment that outlines what we’ve done to track and address safety challenges, including frontier model risks in accordance with our Preparedness Framework. https://t.co/xohhlUquEr
Recommended Articles
— OpenAI (@OpenAI) August 8, 2024
The System Card also contains information on the Preparedness Framework evaluations, which is an essential factor in determining the safety of the AI model in several areas. In particular, the evaluation identified that GPT-4o’s persuasion features presented a borderline medium risk and that OpenAI has implemented measures to reduce these risks.
Joaquin Quiñonero Candela, the head of preparedness at OpenAI, stressed that the company is actively engaged in such research and assessment, especially with regards to practical use.
This is a Breaking Story Please Check Back For More
CoinGape