Anthropic将于2027年上半年开始购买AMD最新一代芯片Instinct MI450
Related
The lack of disclosure of details regarding the Anthropic-US government agreement to lift export restrictions on Fables has sparked controversy.
PANews reported on July 1st that, according to Bloomberg, Anthropic has had its external access restrictions on its Fable 5 and Mythos 5 models lifted after reaching an agreement with the US government. However, the specific agreements between the two parties regarding security standards and technical modifications have not been disclosed. Previously, the ban stemmed from Amazon's feedback to the US government that Fable's security measures could be bypassed. Anthropic subsequently addressed these shortcomings with the support of CAISI, an agency under the US Department of Commerce, and agreed to deepen its cooperation with CAISI, providing computing power and a security team to participate in testing. The report points out that this export control case, reached through private communications and closed-door negotiations, lacks transparency, making it difficult for companies to predict under what standards they might face further blockades in the future.
The EU and Anthropic continue discussions regarding Mythos's visit.
PANews , June 29 - Market sources indicate that the EU and Anthropic continue discussions regarding Mythos's visit.
Anthropic co-founders: Open-source AI carries the risk of irreversible misuse and may lead to a dangerous path.
PANews reported on June 28th that, according to BitcoinNews, Dario Amodei, co-founder and CEO of Anthropic, stated during a speech to lawmakers in the US Congress that the development of open-source artificial intelligence is entering a "very dangerous path." Once powerful AI models are released as open source, developers will lose the ability to effectively monitor model usage, including the inability to monitor abuse, revoke access permissions, or dynamically update security mechanisms, significantly increasing potential risks. Amodei stated that compared to closed model systems, completely open models are more difficult to maintain and control over in terms of security governance, potentially leading to irreversible abuse risks.
OpenAI and Anthropic have maintained a relatively restrained approach to visual technology, leaving room for competition; xAI has launched an upgraded video model, betting on the image generation track.
PANews reported on June 27th that, according to The Information, sources familiar with the matter revealed that xAI is increasing its investment in video and image generation tools, attempting to differentiate itself in multimodal capabilities. In contrast, OpenAI and Anthropic have relatively limited visual generation capabilities, leaving room for xAI to enter the market. It is understood that xAI launched an upgraded video model last week, emphasizing its continued progress in visual generation, while simultaneously introducing external collaborations in areas such as encoding to counter competitors. Sources indicate that xAI is strengthening the user buzz and dissemination effect of its AI video tools, viewing it as a crucial tool for the next stage of product competition. SpaceX has also publicly supported and mentioned the popularity of its AI video tools.
Anthropic's export restrictions remain in effect for two weeks, prompting Asian AI companies to accelerate the rollout of local alternatives.
PANews reported on June 27th, citing TechCrunch, that the US export restrictions on Anthropic's high-end AI models have been in place for two weeks, prompting Asian AI companies to accelerate the development of local alternatives due to overseas supply shortages. Chinese cybersecurity company 360 launched TuLongFeng, a vulnerability discovery AI tool, to compete with overseas products like the Mythos series. Japanese AI startup Sakana AI released its new model, Fugu, whose comprehensive capabilities rival Fable 5 and Mythos Preview, leveraging a unified API to achieve multi-model collaborative scheduling and automatic AI agent orchestration architecture.
The United States allows trusted partners to use Anthropic's Mythos 5 AI model.
PANews reported on June 27 that, according to Cailian Press, the U.S. Department of Commerce stated in a letter that the U.S. has eased restrictions on the use of Anthropic PBC's high-end Mythos 5 AI model by "trusted partners." This move eases the dispute between Anthropic and the U.S. government. Previously, the U.S. government, citing security concerns, abruptly banned Anthropic from providing Fable 5 and Mythos 5 to foreign citizens. The letter, dated Friday, did not mention any adjustments to the restrictions on the use of the Fable 5 model. U.S. Department of Commerce spokesperson Benno Kass stated, "In just two weeks, we have worked diligently to ensure that the United States continues to maintain its leadership in the global field of artificial intelligence while upholding national security."