Anthropic accuses Chinese companies Moonshot and DeepSeek of copying Claude's capabilities
AI News
Искусственный интеллект
Инвестиции

Anthropic accuses Chinese companies Moonshot and DeepSeek of copying Claude's capabilities

SAPO Notícias11 September 2026 at 03:55

Anthropic, a North American artificial intelligence company, accused Chinese AI companies, namely Moonshot AI and DeepSeek, of having used millions of interactions with the Claude model to try to replicate their capabilities through a technique known as "distillation."

The distillation technique consists of using an existing model to generate training data that allows the development of a new model with similar capabilities, without the need to invest the same level of computational resources.

The accusation comes in the context of growing technological competition between the United States and China in the field of artificial intelligence, a sector considered strategic by both powers.

Related articles

AI News

How to prevent Artificial Intelligence from killing us all?

A former Anthropic researcher warns of extreme risks associated with artificial intelligence, while Adolfo Mesquita Nunes explores the extent to which this threat can reach and what measures can still be implemented to control and regulate the development of AI before it is too late.

Observador11/09/26, 05:02
AI News

What Venice-acclaimed documentary reveals about secret system used by Israel against civilians in Gaza

The Israeli documentary "NAZA", acclaimed at the Venice Film Festival, reveals through testimonies from 24 members of the Israeli Defense Forces how secret intelligence and surveillance systems, including artificial intelligence, were used during the war in Gaza to identify, track, and attack Palestinians, including low-ranking Hamas targets. The accounts describe the use of AI to locate people, track cell phones, and intercept conversations between targets and their families before attacks that frequently resulted in the death of all occupants of the houses.

oglobo11/09/26, 04:00
AI News

Productivity that can destroy skills

The article addresses a productivity paradox in the artificial intelligence era: if AI takes over tasks traditionally performed by junior professionals, how will these workers develop the skills needed to become seniors in the future? The text questions the impact of automation on building organizational knowledge, transferring skills, and career development, warning that the pursuit of immediate efficiency may compromise the formation of experienced professionals that organizations will need in the long term.

SAPO Notícias11/09/26, 02:10
Anthropic já impediu investigações que se podiam relacionar com armas biológicas
AI News

Anthropic has already blocked investigations that could be related to biological weapons

A report from Anthropic reveals that the artificial intelligence company blocked at least four investigations potentially related to biological weapons, deciding not to disclose them because they could be merely academic studies. The document "Detecting and Combating AI Misuse", published on Thursday, details how the company assessed the cases and chose not to make them public.

SIC Notícias11/09/26, 00:58