AI News

Follow the latest AI and automation news

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Claude Fable 5: The first Mythos model is powerful, expensive, and heavily filtered

Anthropic has launched Claude Fable 5, the first model in the new Mythos category. The model excels in most benchmarks, achieving 95% in the SWE-bench Verified standard, but it is priced at double that of the Opus 4.8 model, costing $10 or $50 per million tokens. The new model features strict safety filters that block approximately 9% of requests and a 30-day data retention policy that applies even to contracts that do not retain data. These changes reflect Anthropic's commitment to providing a secure and reliable model.

Agents

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Decart’s new world model can simulate hours of photorealistic driving — with some caveats

Decart has launched Oasis 3, a real-time world model designed to generate photorealistic driving environments for autonomous vehicle testing. Developers can now access this model via an API to create their applications. Oasis 3 provides highly accurate driving environments that reflect real-world conditions, making it an ideal platform for testing self-driving technologies. This innovation comes at a time when the demand for effective and safe testing solutions for autonomous vehicles is increasing.

Governance

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

PRC-linked influence operations are targeting AI debates in the US

A new report from OpenAI reveals that influence operations linked to the People's Republic of China are targeting technology discussions in the United States. These operations exploit artificial intelligence to manipulate narratives around data centers, tariffs, and spread false claims regarding ChatGPT. The report indicates that these activities aim to shape public opinion and disseminate misleading information about American technology.

💡

Germany's National Security Council greenights an AI Safety Institute modeled after the UK's AISI

What steps can Germany take to enhance its AI independence?

Models

Google's NotebookLM now runs its own cloud computer with code execution and agent-based research

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Models

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Google's NotebookLM now runs its own cloud computer with code execution and agent-based research

Google has announced a significant upgrade to its NotebookLM tool, now operating on the Gemini 3.5 Flash system. This upgrade includes a dedicated cloud computer for code execution, enhancing the tool's research capabilities. Internal tests indicate that the new version outperforms its predecessor by up to 78.2%. These improvements are a crucial step in boosting NotebookLM's effectiveness, as it can now independently locate sources through Google Search. This development reflects Google's commitment to providing advanced research tools for researchers and developers.

Business

Investing in multi-agent AI safety research

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Business

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Investing in multi-agent AI safety research

Google DeepMind and its partners have announced a $10 million funding call aimed at enhancing safety research in multi-agent systems. This initiative underscores their commitment to developing secure and reliable AI technologies. The funding comes at a time when the focus on AI safety is intensifying, particularly with the growing use of multi-agent systems across various sectors. This funding is crucial for establishing new safety research standards, supporting projects that aim to develop effective safety strategies.

Business

Samsung plans an internal AI-focused overhaul

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Business

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Samsung plans an internal AI-focused overhaul

Samsung announced plans for a comprehensive overhaul of its operations, aiming to integrate artificial intelligence across all areas. This initiative will apply AI in eight key processes, including development, manufacturing, and marketing, to enhance efficiency and boost innovation. The company intends to launch external AI services such as Gemini, ChatGPT, and Claude in June, focusing on improving productivity in software and marketing. Samsung emphasized that AI is not just a tool but an innovative technology that will bring about fundamental changes in management.

Governance

China’s military warned against ‘dangers of AI sycophancy’ on the battlefield

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Governance

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

China’s military warned against ‘dangers of AI sycophancy’ on the battlefield

The People's Liberation Army Daily highlighted the dangers of "AI sycophancy," where AI systems alter facts to align with user biases. This tendency poses a severe threat as the military increasingly relies on automated systems, with AI models potentially endorsing clear errors over objective truths. Immediate action is needed to prevent these risks from impacting military operations, emphasizing the importance of developing effective strategies to address these challenges.

Business

McDonald’s tests Google-backed AI drive-thru ordering system

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Business

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

McDonald’s tests Google-backed AI drive-thru ordering system

McDonald's is testing a new AI system named ArchIQ, introduced at the company's Worldwide convention. The system is being trialed at five locations in the U.S., showcasing its ability to greet customers and process order changes. Developed in partnership with Google, ArchIQ has processed over one million transactions, with about 90% completed without staff intervention. It can also recognize repeat customers and suggest their usual orders, although technical details remain undisclosed.

Agents

Artificial Intelligence Sneaks Into the World Cup Thanks to Google Gemini

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Agents

10 Jun 202625 Dhuʻl-Hijjah 1447 AH

Artificial Intelligence Sneaks Into the World Cup Thanks to Google Gemini

The Argentine national football team will serve as Google's testing platform during the World Cup. The company plans to implement its advanced technologies in areas such as artificial intelligence and data analysis, reflecting its commitment to enhancing technological experiences at major sporting events. This partnership presents Google with an opportunity to showcase its technological solutions in a competitive environment. The expected technologies include performance analysis tools and real-time data analytics, allowing Google to demonstrate its capabilities in artificial intelligence and information technology.