OpenAI, Anthropic and Google DeepMind have announced they are working on the creation of an oversight body to manage risks associated with Artificial Intelligence, in an initiative designed to strengthen controls over the development of advanced AI systems.
The proposal originated with an idea put forward by Demis Hassabis, Chief Executive of Google DeepMind, who suggested establishing a structure modelled on the Financial Industry Regulatory Authority (FINRA), the self-regulatory body for the US financial sector. The proposal envisages an organisation funded by AI companies themselves but governed by a board composed of independent experts and representatives from different segments of the industry.
Under the proposal, the most advanced AI models would be required to undergo safety evaluations before being made commercially available. The body could also coordinate a potential slowdown in the pace of technology development should risk-assessment mechanisms fail to keep pace with the speed at which systems are evolving.
The initiative comes against a backdrop of growing debate over the risks associated with Artificial Intelligence. In recent weeks, the sector has returned to the spotlight following incidents involving AI models and warnings issued by researchers and executives about potential dangers linked to the development of increasingly powerful systems.
Despite the intention to create an industry-internal oversight mechanism, the proposal raises legal questions. Experts cited by the France-Presse news agency warn that an entity of this kind would require a legal framework to avoid accusations of collusion and potential violations of competition rules.
FINRA, used as the reference model for the proposal, holds statutory authority and operates under the supervision of the United States Securities and Exchange Commission (SEC). The organisation can set rules for its sector, but these are subject to SEC approval and oversight.
The proposal has also drawn political criticism. US Senator Bernie Sanders has argued for binding international rules on Artificial Intelligence, while some observers have accused major technology companies of seeking to shape regulation of the sector to their own advantage. David Sacks, an adviser to Donald Trump, warned of the risk of creating an AI cartel, arguing that companies should bear individual responsibility for the safety of their own systems.
The discussion is taking place at a moment when the US government is working to define oversight mechanisms for new AI models. In August, the Trump administration launched an evaluation process for new systems that, according to information released, relies on voluntary measures and lacks clear enforcement mechanisms.
Source: Diário Económico
Original article: https://www.diarioeconomico.co.mz/?p=544845
