AI.info

The Pulse — Page 17

Browse The Pulse on AI.info.

• Hugging Face’s Tokenizers v1 reaches 30x faster encoding

  Hugging Face says its Tokenizers v1 release candidate encodes text three to 30 times faster than v0.23 on an Apple M4 Max. The redesign preserves token IDs while replacing regex splitting, repeated allocations and lock contention with SIMD operations, caching and reusable memory.

• Four AI Companies Face Antitrust Suit Over Slowdown Calls

  Subscribers sued Anthropic, OpenAI, Google and SpaceX in federal court, alleging the companies coordinated to slow improvements to competing AI products. The complaint targets public calls by Dario Amodei, Elon Musk, Sam Altman and Demis Hassabis for industry coordination on AI development.

• Researchers Hide 8 Bits in About 50 LLM-Generated Tokens

  An arXiv abstract describes BAM, a feedback-coding method for black-box LLM steganography.

• Google Home Opens Device Controls and History to AI Agents

  Google’s early-access Home MCP server gives compatible AI clients a documented route to inspect smart-home structures, monitor states, run approved device actions and query event history.

• Lovable co-founder cites $600 million revenue pace at HumanX

  Fabian Hedin said Lovable’s revenue had reached $600 million at HumanX, and cited use by two-thirds of Fortune 500 companies.

• U.S. Wins G20 Backing for Light-Touch AI Rules

  The United States persuaded G20 ministers to endorse the Carolina Principles, a framework that limits new AI-specific regulation and favors targeted rules. The agreement puts Washington’s innovation-first approach against calls for stronger safeguards from countries including Canada.

• Cornelis raises $205M to put compute inside AI networks

  Cornelis Networks announces $205 million in funding and introduces Active Compute Fabric for AI and HPC systems. The architecture adds programmable compute to scale-up and scale-out networking, with Qualcomm joining the company at AI Infra Summit.

• Anthropic Weighs New Model as IPO Pressure Mounts

  Anthropic is considering releasing another AI model after OpenAI’s GPT-6 Astra gains ground with enterprise customers. The decision comes as Anthropic weighs safety work, profitability and the timing of its planned public offering.

• RADAR Scores 0.913 AUC Across 146 Abdominal CT Findings

  A Science study evaluates RADAR, a generalist AI model for abdominal CT diagnosis, across 146 findings and multiple clinical settings.

• Cloudflare Defaults to Blocking AI Training and Agent Crawlers

  Cloudflare’s new defaults take effect on September 15, 2026, blocking AI training and agent crawlers on ad-supported pages while continuing to allow search crawlers. The company says the policy is designed to separate search discovery from model training and real-time agent activity.

• Anthropic Finds AI Can Help Build Weapons and Locate Targets

  Anthropic’s Frontier Red Team says current AI systems can assist with intelligence targeting and simulated weapons engineering, while significant physical-world limits remain.

• AI Shrinks the Time Needed to Find and Exploit Vulnerabilities

  A CTech survey of 30 security researchers finds that commercial AI tools are shortening the path from vulnerability discovery to working exploits. Researchers also warn that AI agents, open-source infrastructure and automated security operations are creating new attack routes and costs.

• Z.AI Raises $5 Billion for Models and Self-Training

  China’s Z.AI raises $5 billion through a Hong Kong share placement and zero-coupon convertible bonds. About 60% of the proceeds will fund next-generation models and the company’s fully self-training system.

• US and China Plan AI Safety Talks the White House Says Are Not Scheduled

  Two sources briefed on the planning told Reuters the United States and China are preparing a mid-September dialogue on AI safety, led by Treasury Secretary Scott Bessent. A White House official says no such meeting is currently planned. Proposed topics include monitoring AI-driven cyberattacks and model distillation.

• OpenAI Says Its Agents Have Reached Research-Intern Level

  OpenAI says coding agents can now complete defined research tasks that would take skilled researchers several days. The company reports 3.1 agent-workdays for every human research workday and targets an automated AI researcher by March 2028.

• Strands harness claims 28% lower token costs across six benchmarks

  Strands Agents has released Strands harness, an open-source general-purpose agent that runs locally or on cloud providers and reports lower token use with nearly equal benchmark scores.

• Pangram Puts Colleges in an AI-Detection Arms Race

  About 1,400 North American colleges have purchased Turnitin’s AI detector, while Pangram is gaining attention for its low reported false-positive rate. Professors and universities are split between expanding detection tools and redesigning assignments around process and human review.

• AstroForge puts its Solo AI system in command of a space probe

  AstroForge is preparing to fly Solo, an in-house transformer-based control system, on spacecraft designed for deep-space missions. The company says the system could diagnose and correct anomalies without waiting for instructions from Earth.

• Google Beam Ships to Five New Countries With 18 Partners

  Google is expanding its Beam 3D video communication platform beyond the United States to Canada, the United Kingdom, France, Germany and Japan. The company is also partnering with Industrious to offer Beam trials at selected U.S. workspaces beginning in October.

• IBM Study Finds 60% of Employees Fear AI Skills Erosion

  An IBM Institute for Business Value study finds that 60% of employees worry AI is eroding their skills, with critical thinking cited most often. The survey also shows a gap between what CHROs say workers need and the skills employees prioritize.

• Enveda Raises $311M to Advance AI-Discovered Drugs

  Enveda has raised $311 million in Series E funding at a $2 billion valuation to advance three AI-discovered drug candidates. The Boulder biotech will use the money to move programs for eczema, metabolic health and inflammatory bowel disease into later-stage trials.

• RRSI Paper Describes a Method for Improving Agent Harnesses

  The arXiv abstract reports that RRSI improves performance on both the split used for evolution and five out-of-distribution benchmarks while reducing policy-token use.