In the new version of the Microsoft and OpenAI partnership, claims of achieving AGI (artificial superintelligence) will no longer be one-sided: any «AGI verdict» from OpenAI will only come into effect after confirmation by an independent panel of experts.
In the new version of the Microsoft and OpenAI partnership, claims of achieving AGI (artificial superintelligence) will no longer be one-sided: any «AGI verdict» from OpenAI will only come into effect after confirmation by an independent panel of experts.
After the companies reached an agreement on updated collaboration rules and intellectual property rights, the partnership introduced a procedure for external verification of AGI claims. According to TechRadar, OpenAI retains the right to first declare «AGI achievement,» but the final status now depends on the decision of independent experts.
For Microsoft, this eliminates the risk of a «one-way button,» where an internal partner statement could automatically change the terms of access to technologies, models, and IP.
The updated agreement stipulates that even after independent verification, Microsoft will not lose access to the OpenAI stack: the rights to the models and products will continue until at least 2032. Instead of a corporate declaration with uncertain consequences, the market will receive a formalized procedure, as external experts will have to assess whether the declared level corresponds to the real properties of AGI. This means the ability to solve a wide range of tasks at or above the human level, transfer knowledge between domains, learn autonomously, and demonstrate consistent causal reasoning, not just another «big autocomplete.»
The most questions now concern the methodology: who exactly will form the commission, what will be the mandate and transparency of the selection, what benchmarks will be considered relevant (from multimodal tests to long-term planning tasks), will «red teams» and external audits be allowed, and how will conflicts of interest be avoided. Despite these unknowns, the very fact of independent verification increases predictability for investors, developers, and regulators, reducing the likelihood of high-profile PR statements without technical support.
The concept of AGI is a target threshold for AI capabilities. The industry is moving towards standardization of evaluation: big players are scaling up external verification (security audits, red teaming, evals), and the Microsoft-OpenAI deal is raising the bar — the «AGI announcement moment» is being translated into a process with external validation.
Previously, dev.ua wrote about how OpenAI agreed to transfer a stake worth at least $100 billion to its non-profit management fund and has preliminarily settled financial issues with Microsoft. These are steps that open the way for a large-scale reorganization of the company and a change in its status.
