Are Leading AI Companies Leveraging Safety Concerns to Cement Market Dominance?

Leading Not Only in Model Development but Also in Setting Safety Standards

Safety Protocols Could Become Barriers for New Entrants

Experts Diagnose Efforts to Solidify Market Dominance

Analysts suggest that AI developers' leadership in establishing AI risk verification and evaluation frameworks is a strategy to strengthen market dominance under the banner of safety. The technologies and risk response methods of leading companies may become industry standards, potentially creating entry barriers for latecomers.

Reuters Yonhap News

According to industry sources on September 29, OpenAI and Anthropic are calling for the establishment of independent evaluation frameworks to verify the risk of AI models during development. This approach involves external independent evaluators accessing internal systems and data to assess the riskiness and safety measures of AI models. In this system, the developer sets the audit standards and selects the external evaluation institution.

This trend may strengthen the market dominance of leading companies. The Associated Press analyzed OpenAI and Anthropic's push for increased safety by stating, "It appears they are trying to garner public favor and establish their own safety protocols in a largely unregulated market." The report further noted that this could help build investor trust ahead of initial public offerings and potentially influence the upcoming U.S. midterm elections.

Aidan Gomez, CEO of Cohere, said, "While I support independent AI safety evaluations, the critical issue is who will create the evaluation standards and select the evaluators." He argued, "If a handful of leading companies set the rules, their current market advantage could become entrenched as an entry barrier for newer firms." Kim Myung-joo, head of the AI Safety Research Institute, also commented, "Technical competition ultimately comes down to standards and certifications," adding, "After leading the development of frontier AI, aiming to set standards and certifications as well means striving for technological hegemony."

There are also concerns that capital and resources could become even more concentrated. Harrison Lofthus, Senior Research Analyst at PitchBook, analyzed that "large AI developers may solidify their position as safe investment destinations to secure funding and computing resources, while simultaneously hindering the growth of smaller competitors."

The competition among developers around cybersecurity-specialized models can also be interpreted in a similar context. With concerns growing over cyberattacks using AI, it is being argued that companies are emphasizing security performance to build trust in their own models. Kim Eunseong, head of the Threat Response Policy Team at the Korea Internet & Security Agency (KISA), said, "Highlighting cybersecurity capabilities serves the purpose of associating the image of 'an AI good at security' with the overall safety and trustworthiness of a company's models in order to capture the market lead."

This content was produced with the assistance of AI translation services.