Microsoft-backed (MSFT) AI firm OpenAI (PC:OPAIQ) has released two new AI models (gpt-oss-safeguard-120b and gpt-oss-safeguard-20b) that help developers detect online safety issues. These are improved versions of OpenAI’s earlier gpt-oss models that were released in August. The numbers in their names refer to their size. In addition, OpenAI is offering them as open-weight models, meaning that the internal settings used during their training are publicly available. This gives developers more transparency and control. However, unlike open-source models, the full code can’t be changed or customized.
Elevate Your Investing Strategy:
- Take advantage of TipRanks Premium at 50% off! Unlock powerful investing tools, advanced data, and expert analyst insights to help you invest with confidence.
Interestingly, these models are built to fit different safety policies and are designed to show their reasoning, so developers can understand how each decision was made. For example, a website that hosts product reviews could use the models to flag fake reviews. Similarly, a video game forum could use them to identify posts about cheating. OpenAI worked on the models with ROOST (Robust Open Online Safety Tools), which is a group focused on making AI safer. Meanwhile, Discord and SafetyKit helped test them.
The models are now available as a research preview that can be downloaded on Hugging Face, and OpenAI is asking for feedback from the research and safety communities. Moreover, as part of the launch, ROOST is creating a new community for people working to protect online spaces with AI tools. This move may help OpenAI respond to criticism claiming that it’s growing too fast without paying enough attention to safety.
Is MSFT Stock a Buy?
Turning to Wall Street, analysts have a Strong Buy consensus rating on MSFT stock based on 33 Buys assigned in the last three months. Furthermore, the average MSFT price target of $629.43 per share implies 17% upside potential.
