Anthropic announced on Monday the launch of a new initiative aimed at funding the development of new benchmarks capable of assessing the performance and impact of artificial intelligence models, including its own generative model, Claude.
According to information released in Anthropic's official blog, the company will provide financial support to third-party organizations to develop tools that "effectively measure the advanced capabilities of artificial intelligence models." Interested organizations can submit applications, and evaluations will be conducted on a rolling basis.
Anthropic stated that this investment aims to enhance the entire field of artificial intelligence safety, providing valuable tools to the entire ecosystem. The company believes that developing high-quality, safety-related assessments remains challenging and that demand exceeds supply.
This initiative focuses on artificial intelligence safety and social impact, planning to create challenging benchmarks through new tools, infrastructure, and methods. Anthropic specifically requires testing to evaluate the models' abilities in areas such as cyberattacks, weapon improvement, manipulation, or deception. Additionally, the company is committed to developing a "early warning system" for identifying and assessing artificial intelligence risks related to national security and defense.
Anthropic also indicated that the new plan will support research into the potential of artificial intelligence in assisting scientific research, multilingual communication, reducing bias, and self-regulation. To achieve these goals, the company envisions establishing a new platform where experts can develop assessments and conduct large-scale trials.
While Anthropic's move has been praised, it has also sparked some skepticism. Some views suggest that considering the company's commercial interests, the fairness of its funded projects may be compromised. Moreover, some experts express doubts about certain "catastrophic" and "deceptive" artificial intelligence risks mentioned by Anthropic, fearing that this might divert attention from more pressing current issues of artificial intelligence regulation.
Anthropic hopes that this plan will drive comprehensive artificial intelligence assessment to become an industry standard. However, whether independent artificial intelligence benchmark development groups will be willing to collaborate with commercial artificial intelligence suppliers remains to be seen.