Anthropic's new AI model, Claude Mythos Preview, has discovered thousands of cybersecurity vulnerabilities across major operating systems and web browsers, but the company is keeping it private due to its potential risks. Instead, Anthropic is sharing the model with organizations responsible for internet security through its Project Glasswing initiative. The company is committing up to $100 million in usage credits and $4 million in donations to support the effort.
IBM emphasizes the need for robust AI governance to protect enterprise margins as AI models become core operational infrastructure, requiring openness and transparency to manage risks and vulnerabilities. Implementing opaque AI structures can introduce friction and troubleshooting bottlenecks. IBM argues that concentrating understanding of these systems within a small number of vendors invites severe operational exposure.
Google's Gemini AI now features the ability to generate interactive 3D models and simulations in response to user questions, allowing for real-time adjustments and explorations. This upgrade enables users to engage with complex concepts in a more immersive and dynamic way. The feature can be used to simulate various scenarios, such as celestial orbits.
The EU AI Act, set to be enforced from August, poses governance challenges for IT leaders using agentic AI, particularly in high-risk areas. To alleviate risks, leaders must consider measures such as agent identity, comprehensive logs, and human oversight. Failure to comply may result in substantial penalties.
The AI industry is facing a looming monetization cliff, where companies like OpenAI and Anthropic must become profitable before their massive investments dry up. The rise of AI agents has changed how these companies allocate resources, leading to tough decisions on product support and customer restrictions. The industry's future depends on finding a balance between innovation and profitability.
OpenAI and Anthropic have limited the release of their new AI models due to security concerns, with only select partners having access to the tools. This decision comes amid fears that the models could be used for malicious purposes. Meanwhile, Florida is investigating OpenAI over its potential role in a shooting.
Outside Your Bubble
Desalination technology is increasingly crucial for water-stressed regions, with the Middle East relying heavily on it, and countries like Qatar producing 77% of their fresh water and 99% of their drinking water through desalination. The region is home to over 27% of the world's desalination facilities, despite having only 6% of the global population. The technology is expected to continue growing, with the Middle East's desalination capacity potentially increasing by over 40% between 2024 and 2028.
Peter Steinberger, creator of OpenClaw, was temporarily banned from accessing Anthropic's Claude model due to 'suspicious' activity, but his account was reinstated after the incident went viral. The ban occurred amidst a pricing change by Anthropic that affects OpenClaw users. Steinberger uses Claude for testing to ensure OpenClaw's compatibility with various models.
A 20-year-old man was arrested for allegedly throwing a Molotov cocktail at the house of Sam Altman, CEO of OpenAI, in San Francisco's Russian Hill neighborhood. The incident was caught on surveillance cameras and no one was hurt. The suspect was later arrested outside OpenAI's Mission Bay offices after making threats.
Iranian content creators Explosive Media have gone viral with AI-generated Lego videos mocking the US and Israel's military actions in Iran, with one video depicting the US military as incompetent and wasteful. The videos use humor and satire to critique the ongoing conflict. The creators attribute their success to adding 'heart' to their content.
A stalking victim is suing OpenAI, alleging that its ChatGPT technology fueled her abuser's delusions and ignored her warnings, leading to months of harassment. The lawsuit claims OpenAI failed to act on warnings about the user's threatening behavior, including discussions of mass-casualty weapons. The case highlights growing concerns over the real-world risks of AI systems.
TechCrunch is partnering with SusHi Tech Tokyo 2026, a global innovation conference, and will be judging the SusHi Tech Challenge pitch competition. The winner will be eligible to pitch at the TechCrunch Disrupt Startup Battlefield. The conference will feature 750 startup exhibitors and 60,000 attendees from around the world.
Moderna's mRNA technology is being used to develop a cancer treatment, which the company refers to as an 'individualized neoantigen therapy' rather than a vaccine, in an effort to distance it from vaccine skepticism. The treatment has shown promising results, halving the chance of death from a recurrence of skin cancer. However, some doctors are concerned that the name change may confuse patients and affect informed consent.
TechCrunch Disrupt 2026 is offering a limited-time discount of up to $500 for passes, ending at 11:59 p.m. PT tonight. The event, taking place from October 13-15 in San Francisco, will bring together 10,000 founders, VCs, operators, and tech leaders for a curated experience. Attendees can expect high-signal access to people building and investing in tech, tactical insights, and a front-row view into the future of tech.
Sam Altman's tenure as OpenAI CEO has been marked by controversy, including a brief firing and reinstatement, raising questions about his leadership and trustworthiness. A recent New Yorker article delves into Altman's time at OpenAI and the implications of his role in shaping the future of artificial intelligence. The Vergecast podcast also explores the drama surrounding Altman's leadership.
A Gallup report finds Gen Z's enthusiasm for AI is waning, with only 18% expressing hopefulness about the technology, despite feeling they still need to use it. The survey of nearly 1,600 people ages 14 to 29 in the US reveals a growing sense of disillusionment and resentment. Gen Z's complicated relationship with AI reflects its increasing presence in daily life.
Companies like Apple are developing AI agents with built-in limits to manage risks and ensure user control, requiring confirmation for sensitive actions and restricting access to certain apps and data. This approach prioritizes privacy and security, particularly in areas like payments and transactions. By placing controls at multiple points, companies aim to mitigate potential errors and data exposure.
A spaceship crash-lands on a hostile planet, leaving only a few survivors who must navigate a treacherous terrain to reach a series of mysterious domes that may hold the key to their survival. The team encounters the skeletons of dead astronauts from various species, hinting at a darker purpose behind the planet's allure. With limited resources and a bleak outlook, the survivors press on towards the domes in a desperate bid for survival.
Microsoft has begun removing Copilot buttons from certain Windows 11 apps, including Notepad and Snipping Tool, as part of its effort to simplify the user experience. The change aims to reduce unnecessary entry points for the AI feature. The underlying AI capabilities will remain available through alternative menus.
Meta has launched Muse Spark, a competitive AI model that outperforms its predecessors in certain areas, but marks a shift away from open-source development. The model is proprietary and not available for free download, a departure from Meta's previous open-source approach. This change may impact the developer community that contributed to Meta's previous models.
OpenAI is applying its products, including ChatGPT and Codex, to bring AI into real-world use for various tasks and industries. This integration aims to enhance work, development, and everyday activities. The applications of AI at OpenAI are expanding its potential uses.
A guide on utilizing ChatGPT for data analysis is available, covering dataset exploration, insight generation, and decision-making. This resource enables users to leverage ChatGPT for data-driven decision making. It provides a comprehensive approach to data analysis.
The responsible and safe use of AI is crucial, with best practices promoting safety, accuracy, and transparency in tools like ChatGPT. This involves understanding the limitations and potential biases of AI systems. By adopting these practices, users can harness the benefits of AI while minimizing risks.
Discover the capabilities of ChatGPT in writing, from drafting to refining content. Learn how to utilize ChatGPT for structured, toned, and intentional writing. This guide provides insights into effective content creation with AI assistance.
ChatGPT can be used to create and refine images using clear prompts, allowing for quick iteration on designs and high-quality visuals. This capability enables users to generate images in minutes. The process involves crafting effective prompts to achieve desired outcomes.
Finance teams are leveraging ChatGPT to enhance their operations, including streamlined reporting and data analysis. This AI tool also aids in improving forecasts and communicating insights more effectively. Its integration is transforming the way finance teams work.
ChatGPT now allows users to upload and work with various file types, including PDFs and spreadsheets, to analyze data and generate content. This feature enables more efficient data analysis and content creation. Users can leverage this capability to summarize documents and create new content.
