Anthropic deploys Claude Sonnet 5, Fable and Mythos restored

artificialintelligence-news.com Jul 1, 2026

Anthropic deploys Claude Sonnet 5, Fable and Mythos restored

Curated by Shubham · TechCapsules · Jul 1, 2026

AI-summarised brief · reviewed before publication

Anthropic has deployed Claude Sonnet 5, a high-capability AI model, and restored access to its Fable and Mythos frontier models following a federal export control review. The decision concludes an 18-day operational pause triggered by a US government directive on June 12, which was enacted after researchers at Amazon documented a method to bypass safety controls of Fable 5. Anthropic has since developed an updated automated classifier to patch the vulnerability, clearing the path for a full commercial rollout.

💡 Why It Matters

· The regulatory pressures facing frontier intelligence systems underscore the need for robust safety measures, as highlighted by the vulnerability identification behavior in Fable 5.
· The development of an automated safety classifier demonstrates Anthropic's commitment to addressing these concerns and ensuring the secure deployment of its high-capacity models.

Read full article on artificialintelligence-news.com

NVIDIA BioNeMo accelerates Anthropic Claude Science

Anthropic Claude Science has integrated the NVIDIA BioNeMo Agent Toolkit to accelerate computational life sciences research. The platform enables scientists to converse with digital agents using natural language to execute research workflows, connecting natively to NVIDIA's toolkit and exposing high-performance computing resources. This integration imports NVIDIA-accelerated models and libraries into the environment, allowing researchers to run sophisticated workflows and increase iteration speeds. The system translates natural language intent into operational action, orchestrating execution using preconfigured agents. Top pharmaceutical companies have already deployed NVIDIA BioNeMo, demonstrating its high penetration across the ecosystem, with Claude Science facilitating fast iterative loops between human reasoning and machine-accelerated processing.

💡 By streamlining complex molecular design workflows, scientists can focus on core science, driving breakthroughs in fields like cancer research. Accelerated workflows and advanced open models enable autonomous agents to reason and complete tasks at maximum hardware speed.

windowslatest.com · Jul 2

Microsoft Teams’ new controversial AI will listen to your meetings and answer…

Microsoft Teams is introducing a new AI-powered feature called Facilitator, which can detect knowledge gaps during meetings and proactively address them by sharing relevant answers. The feature, which will be disabled by default, uses web search in chat to provide responses. Admins can control where the feature is enabled, and users with a Microsoft 365 Copilot Premium license must manually add it to meetings. The feature is expected to reduce interruptions and make meetings more productive.

💡 Microsoft's use of AI to monitor and respond to meeting conversations raises concerns about data privacy and the potential for AI to misinterpret or manipulate meeting content.

neowin.net · Jul 2

Copilot in Excel is getting improved web search this month

Microsoft is enhancing the web search capabilities of Copilot in Excel, a feature that allows users to search the web for answers within the application. A new multi-agent web search system is being introduced, which will coordinate parallel research, verify findings, and re-check gaps when needed. This improvement aims to provide better-sourced answers and context, as well as reduce hallucinations. The rollout is planned for this month and will be available to users with a commercial Copilot and Microsoft 365 license.

💡 Improved web search capabilities in Copilot will make Excel a more reliable tool for users seeking accurate information, potentially reducing the risk of spreading misinformation.

easternherald.com · Jul 2

Apple Built an AI Agent Interface Into Safari, and It Stays Completely…

Apple has integrated an AI agent interface into Safari, utilizing the Model Context Protocol, allowing AI agents to navigate and debug within the browser without Apple's involvement. This native implementation enables AI agents to connect directly to a Safari browser window, taking screenshots, reading the DOM, and running JavaScript. Safari is the first major browser to support this protocol, providing a consistent way for AI agents to interact with the browser's internals. The update aims to streamline the workflow of AI-assisted web development, reducing friction and handoffs between developers and AI assistants. Apple's WebKit team released Safari Technology Preview 247 with this feature.

💡 By giving AI agents direct access to browser internals, developers can automate testing and debugging tasks more efficiently. This integration also enables AI assistants to provide more accurate and effective support, as they can now interact with the browser directly.

windowslatest.com · Jul 2

Microsoft brags Copilot key has main character energy on Windows 11, but…

Microsoft recently posted a social media ad calling the Copilot key a button with "main character energy," despite admitting it hurts productivity for some users. The company will allow users to remap the key to act as Right Ctrl or the Context menu key again through a Windows 11 update later this year. The ad was met with negative feedback, with many users expressing their dislike for the Copilot key and asking for its removal. Microsoft had previously confirmed the key caused disruption to productivity and accessibility workflows. The company's decision to let users remap the key is seen as a fix for the issues it caused. The Copilot key has been a point of contention for Microsoft, with many users finding it unnecessary and intrusive.

💡 Microsoft's attempt to rebrand the Copilot key as a desirable feature backfired, highlighting the company's misstep in pushing the technology on users. The backlash against the Copilot key undermines Microsoft's efforts to promote its AI capabilities as a key selling point.

techcrunch.com · Jul 2

Yep, we’re using OpenClaw to date now

Ben Guez, a content creator and startup founder, has used OpenClaw, an open-source AI agent, to automate his dating life on Instagram. He uses the AI to track World Cup match results and then creates and posts Instagram "trial reels" with a template, garnering over one million views and 200 DMs in a few days. Guez's method involves posting a video with a caption expressing emotional support for women from countries that lost a match, and he requires them to download his AI language learning app to respond. Other individuals, such as Jeff Weisbein, are also using OpenClaw to streamline their dating processes, albeit in less unconventional ways. The use of AI in dating is raising questions about the boundaries of automation in personal relationships. Guez believes that as long as he is open about his methods, it is fine, but others may feel played or impressed by his creativity. The reactions of the women involved are varied, with some being impressed and others potentially feeling angry.

💡 The use of AI in dating highlights the blurring of lines between human connection and automation, raising questions about what is acceptable in personal relationships. Delegating emotional labor to AI agents challenges traditional notions of intimacy and communication.