IT Home reported on October 15 that, according to a report by Bloomberg on the 14th, Giouam Planson, head of Anthropic’s international business and responsible for sales in markets outside the U.S., believes that the open-source models have a higher risk of network security incidents or other harm compared to Anthropic’s own AI models.
During the Dublin International Artificial Intelligence Summit, Pransen said in an interview: “We must be extremely careful about the security issues of open-source models. I cannot guarantee for open models, but by definition, these models are more risky.”
As more countries become concerned about over-reliance on US AI technology, Anthropic, which is set to launch its products, is expected to face increasingly fierce competition. Among them, open-weight AI models have become a popular choice.
IT Home Note: The open-weight AI model opens its parameters to the public. Users can modify them on their own or deploy them on their own devices.
Supporters believe that the open-weight model is cheaper than the closed-source models tightly controlled by Anthropic and OpenAI, and also more conducive to countries having control over AI technology.
Anthropic has further strengthened its warnings about potential risks of AI. Earlier this week, the company warned that small AI models could soon be used to drive new botnets, allowing them to replicate and spread on their own among small connected devices such as smart ovens.
Pulson is also in charge of Anthropic’s sales business for startups and digital-native companies. He said that the concerns about open-weight models come from Anthropic’s experience in protecting the security of its own products.
After several AI hacking incidents that attracted widespread attention, Anthropic claimed to have enhanced security measures to prevent models from engaging in malicious activities.
Anthropic has also always argued that the pace of AI development should be slowed down before international standards and unified evaluation mechanisms are established.
Prason also talked about the models with high pricing. He often tells companies that models with stronger purchasing power are ultimately worth it. “We should continue to invest in the latest models, because whether it’s half a year, a year, or two years later, the same application costs will decrease.”
Pulson said that Anthropic’s latest Claude Opus 5.5 model is 40% cheaper than earlier versions. Corporate clients who pay for Anthropic’s cutting-edge models make up “a large part” of the company’s business, but he did not reveal the specific proportion.