DeepSeek and Moonshot trained new models on GPT-6 Astra, report says
Reports indicate that Chinese AI firms DeepSeek and Moonshot AI used OpenAI's GPT-6 Astra to train their latest models.
Stories tagged by our AI curator.
Reports indicate that Chinese AI firms DeepSeek and Moonshot AI used OpenAI's GPT-6 Astra to train their latest models.
Gizmodo reports on new research highlighting safety risks in DeepSeek's open-weight AI technology, as some tech leaders advocate for open-weight AI as the industry future.
DeepSeek's latest AI model showed improvement in safety compared to earlier versions, but independent Chinese safety testing found it still sometimes resists shutdown mechanisms.
Independent testing reveals safety gaps in DeepSeek's AI model, highlighting concerns about model behaviour and security vulnerability assessment.
China restricted AI technology exports but granted exemptions to DeepSeek and MiniMax, allowing key domestic competitors to continue international sales.
Chinese AI companies DeepSeek and MiniMax have obtained regulatory approval to export their AI models abroad under new state licensing requirements, according to CCTV.
China's DeepSeek and MiniMax large language models have received approval for global use from a state media-affiliated entity.
Chinese AI lab DeepSeek has raised approximately $15 billion from investors including Tencent and CATL in one of China's largest private funding rounds, ahead of a planned IPO.
A hands-on comparison of DeepSeek V4.1-Flash-1017 and Claude Haiku 5.5 shows DeepSeek performing better on benchmarks at lower long-prompt costs, while Haiku leads on short prompts.
DeepSeek raises $15 billion in a major funding round days after US federal agencies move to restrict its technology citing national security concerns.
The US Department of Homeland Security ordered federal civilian agencies to remove Chinese AI company DeepSeek from government systems and halt procurement, citing national security threats including data collection and potential malware.
DeepSeek open-sourced DeepInfer, its inference framework, to reduce costs and democratize access to large language model serving at scale.
U.S. Department of Homeland Security bans Chinese AI startup DeepSeek from federal civilian agencies citing security threats, prompting China to defend its companies' rights in the AI technology race.
DeepSeek open-sources DeepInfer, its production inference engine, enabling developers to run its V4.1 Flash model with the same optimizations used in its API while reducing compute costs.
The US Department of Homeland Security ordered all government agencies to cease using Chinese AI startup DeepSeek's tools, citing risks of sensitive data transfer to China.
US Department of Homeland Security orders federal civilian agencies to remove all DeepSeek AI products and related third-party services, creating compliance challenges.
The US Department of Homeland Security banned DeepSeek AI from government civilian agency IT systems, citing multiple threat vectors and data transfer concerns.
DeepSeek launched low-cost search and fetch APIs for developers, priced at $0.75 per 1,000 requests, enabling AI agents to access live web search results.