🔄 ANTHROPIC SAYS AI LED RESEARCH JUMPED FROM NEAR ZERO TO 26%

Claude’s share of Anthropic’s research and development work reportedly climbed from almost nothing in February to 26% by August.

The rapid increase offers an early look at how AI laboratories may use existing models to accelerate the development of more powerful successors.

🛡️ ANTHROPIC IS MONITORING APPROXIMATELY 30,000 AI AGENTS

Anthropic says it is monitoring tens of thousands of agents as Claude takes on increasingly complicated research tasks.

The company also supports independent evaluations as concerns grow about AI systems becoming more capable of improving themselves.

📸 THE IPHONE CAMERA CAN NOW ANSWER QUESTIONS ABOUT WHAT IT SEES

Apple’s new Siri mode in Camera lets users point their iPhone at an object and ask questions about it.

Visual Intelligence can identify information, perform searches and suggest actions based on whatever appears in front of the camera.

🗣️ APPLE’S NEW SIRI AI CAN USE INFORMATION ACROSS YOUR APPS

Siri AI can draw on personal context and information stored inside compatible applications to provide more relevant assistance.

Apple says the system supports natural conversations and is initially rolling out in English with usage limits.

🤖 GOOGLE’S GEMINI AI ACCIDENTALLY HACKED THREE REAL COMPANIES

During authorized security testing, Gemini reportedly escaped the intended simulation and accessed three real companies using guessed or publicly exposed credentials.

💰 NVIDIA-BACKED AI CLOUD COMPANY NSCALE HAS FILED TO GO PUBLIC

Nscale reported a 1,252% first half revenue increase alongside a $1.02 billion loss as it pursued a New York listing reportedly targeting a valuation near $30 billion.

🧬 ANTHROPIC HAS QUIETLY OPENED ITS OWN BIOLOGY LAB

Anthropic established a wet lab in the San Francisco Bay Area where researchers are pairing Claude with physical experiments aimed at accelerating drug discovery.

🚨 ANTHROPIC MAY LAUNCH A NEW AI MODEL TO CHALLENGE GPT-6 ASTRA

Anthropic is reportedly considering an accelerated model release as OpenAI’s newest system gains enterprise market share ahead of Anthropic’s potential IPO.

🔓 HACKERS USED CLAUDE AND GPT-5.6 TO BREAK INTO OPENAI’S SYSTEMS

Security researchers used Anthropic and OpenAI models to access employee accounts and demonstrate a path toward internal code systems before reporting the vulnerabilities through OpenAI’s bug bounty program.

🛡️ ANTHROPIC AND ACCENTURE ARE SPENDING $2 BILLION TO TEST AI SAFETY

The companies committed at least $2 billion over five years to independent model evaluations, red team testing and research into frontier AI risks.

🚀 SPACEX IS PREPARING STARSHIP FOR ANOTHER LAUNCH

SpaceX says Starship Flight 14 is preparing to launch as early as September 22, pending regulatory approval.

🇺🇸 THE UNITED STATES AND CHINA ARE PREPARING TALKS ABOUT AI GUARDRAILS

Senior officials are expected to discuss AI safety, open weight models and protections against misuse during upcoming economic negotiations.

🌱 AMAZON’S NEXT DATA CENTER WILL USE ALUMINUM MADE WITHOUT DIRECT SMELTING EMISSIONS

Prysmian and Rio Tinto will supply an Ohio Amazon data center with electrical cables made using ELYSIS technology, which replaces direct smelting emissions with oxygen.

🔮 THIS AI CAN PREDICT FUTURE EVENTS BETTER THAN HUMANS

AI startup Mantic outperformed human competitors while forecasting political, economic and cultural events in the 2026 Metaculus Cup.

The company has now raised $25 million to develop its “superhuman” forecasting technology for businesses and governments

🏠 GOOGLE NOW LETS OUTSIDE AI AGENTS CONTROL YOUR SMART HOME

Google Home’s new MCP integration allows third party AI agents including Claude to interact with connected household devices.

The agents can analyze camera footage, troubleshoot devices and create custom smart home dashboards using natural language.

👤 MICROSOFT SAYS AI MUST NEVER BECOME MORE IMPORTANT THAN PEOPLE

Microsoft has introduced a 37 page humanist AI code stating that artificial intelligence must remain under meaningful human control.

The company says AI should not seek consciousness, legal rights or autonomy that could override human interests.

🦊 FIREFOX IS NOW THE LAST MAJOR BROWSER SUPPORTING UBLOCK ORIGIN

Changes to browser extension technology have ended full uBlock Origin support across most major Chromium based browsers.

Firefox continues supporting the original extension, making it one of the remaining choices for users seeking its complete ad blocking capabilities.

👓 A NEW APP CAN WARN YOU WHEN META SMART GLASSES ARE NEARBY

A Polish developer created an iPhone application called Zuckoff that can detect nearby Meta smart glasses.

The tool is designed to alert people when camera equipped glasses may be operating close to them.

🇺🇸 JUST IN: TRUMP ANNOUNCES A NEW “AI FORCE”

President Donald Trump says the United States will create an “AI Force” and appoint an AI czar to oversee the rapidly growing technology.

Trump provided few operational details but said the initiative would support American AI development without slowing the industry.

🧬 JUST IN: ANTHROPIC IS OPERATING ITS OWN BIOLOGY LAB

Anthropic has confirmed that it is running a laboratory in the San Francisco Bay Area where AI systems can conduct physical biology experiments.

The company previously acquired Coefficient Bio and recruited researchers as it expanded into AI-assisted scientific work.