AI Daily Digest
Your daily briefing on AI developments that matter to your business.
🎧 Listen to this digestWednesday, September 23, 2026
Government & Policy
2Trump Proposes Renaming 'Artificial Intelligence' to 'Super Intelligence' at UN General Assembly
Speaking at the UN General Assembly on September 22, President Trump proposed that the U.S. government replace the term 'artificial intelligence' with 'super intelligence' in official documents, framing it as a branding shift to reflect the technology's power. No formal executive order or binding federal directive has been confirmed to implement the change, and no other nation or the UN itself has indicated it will adopt the new terminology. The proposal has no immediate impact on existing AI regulations, procurement rules, or compliance requirements.
Gadget ReviewSen. Bernie Sanders and Rep. Greg Casar Introduce Bill to Ban Artificial Superintelligence and Create Department of AI
Progressive lawmakers Senator Bernie Sanders and Representative Greg Casar unveiled legislation on September 23 that would ban the development of artificial superintelligence and establish a new federal Department of AI. The bill represents the most aggressive legislative proposal yet from the left flank of Congress on AI governance, going significantly further than existing state-level measures. The legislation faces an uphill battle given the current administration's pro-development stance, but signals growing bipartisan interest in structural AI oversight.
KING5 NewsMajor AI Players
3OpenAI Launches GPT-6 Sol and GPT-6 Luna; Anthropic Releases Claude Opus 5.5 with 40% Lower Costs
On September 22, OpenAI released two new frontier models — GPT-6 Sol and GPT-6 Luna — just 18 days after GPT-6 Astra, continuing a pace of one major flagship model every 17 days industry-wide. Simultaneously, Anthropic launched Claude Opus 5.5, touting 40% lower running costs than its predecessor alongside stronger performance and enhanced safety safeguards. The dual launch day underscores how rapidly the competitive landscape is shifting, with brands and businesses facing model changes that can alter AI recommendations and outputs on a near-monthly basis.
SearchIntelOpenAI CEO Sam Altman and Anthropic's Dario Amodei Address UN Security Council on AI Safety
France, holding the September UN Security Council presidency, convened a historic 15-member session on AI and international security on September 23, with OpenAI's Sam Altman and senior Anthropic representatives expected to address world leaders alongside representatives from Chinese labs including DeepSeek. The session coincides with growing calls for international AI governance frameworks amid rapid capability advances. It marks the first time major AI lab CEOs have addressed the Security Council directly on safety and regulation.
AI WeeklyAnthropic Files for IPO at Expected $2 Trillion Valuation — Potentially the Largest in History
Anthropic confidentially submitted a draft S-1 registration to the SEC in June 2026, with investors now expecting it to float at a valuation of $2 trillion or more in October — which would surpass SpaceX's $1.77 trillion IPO in June and set a new all-time record. The five-year-old company's IPO prospectus will face scrutiny over the sustainability of AI revenue growth assumptions. OpenAI and Anthropic alone absorbed approximately 43% of global venture capital in the first half of 2026.
EmpowerGlobal Developments
1UK Declines to Sign 21-Nation AI Oversight Declaration; Trump Rejects 'Globalist' AI Control at UN
A new development in the 21-nation AI oversight push: the UK, despite Prime Minister Andy Burnham's stated ambition to broker a global AI agreement during Britain's G20 presidency, did not sign the declaration published by the Dutch government — joining the US, China, France, Japan, and India on the sidelines. At the UN General Assembly, Trump explicitly rejected 'any attempt to construct a globalist scheme' to control AI and dismissed warnings of AI existential risk as 'a hoax.' Former UK cabinet minister Darren Jones publicly called on Burnham to sign and help broker a US implementation agreement, increasing domestic political pressure.
BBC NewsTransportation
1Mercedes-Benz Partners with UK Startup Wayve to Bring Tesla FSD-Like Driving Assistant to Production Cars Within Two Years
Mercedes-Benz has signed a deal with Wayve, a UK-based autonomous driving startup, to integrate Wayve's 'AI Driver' technology into upcoming Mercedes vehicles, enabling advanced urban point-to-point driving assistance similar to Tesla's Full Self-Driving feature. The rollout is targeted within two years, as legacy automakers increasingly race to match Tesla's end-to-end AI driving system. Other automakers including Rivian, Toyota, Hyundai, and Lucid are pursuing similar Level 2++ capabilities through partnerships with Nvidia and proprietary systems.
Inkl / Automotive NewsHealthcare & Biotech
1AACR Annual Cancer Report Highlights AI-Assisted Screening, Early Detection, and FDA-Approved AI Diagnostic Tools
The American Association for Cancer Research's 2026 annual progress report highlights AI as a transformative force in cancer care, with the FDA having approved multiple AI-based tools for early cancer detection and diagnosis over the past year. AI is being used to improve interpretation of routine medical images, enable high-resolution spatial mapping of tumor microenvironments, and accelerate discovery of new cancer treatments. The report frames AI not as a future promise but as a present clinical reality with measurable patient outcomes.
ASCO PostInvestment Noteworthy
2AI Drug Discovery Startup Enveda Raises $311 Million, Doubling Valuation to $2 Billion
Boulder-based Enveda, which uses AI to discover drug compounds found in nature, raised $311 million in a new funding round that doubled its valuation to $2 billion. The company plans to use the capital to advance three clinical-stage candidates into later-stage trials for inflammatory and metabolic conditions, while expanding its AI platform. Enveda's approach — letting AI identify naturally occurring molecules rather than synthesizing novel ones — represents a growing sub-segment of the AI drug discovery market.
QuartzSouth Korea's AI Unicorn Wrtn Technologies Selects Banks for 2027 IPO
South Korean AI services startup Wrtn Technologies, which achieved unicorn status after a 100 billion won Series C in August 2026, has tapped Mirae Asset and NH Investment & Securities to lead its planned IPO next year. Wrtn is the only pure AI services company among South Korea's five AI unicorns, with overseas revenue now surpassing domestic revenue. The listing would be a bellwether for the health of the AI services sector in Asia's public markets.
BigGo FinanceGeneral AI
4UC Berkeley's GPN-Star AI Model Sets New Standard for Identifying Disease-Causing Genetic Variants
Researchers at UC Berkeley developed GPN-Star, a genomic language model that significantly outperforms existing tools at identifying genetic variants that cause inherited diseases by learning the 'grammar' of the genome. The team published genome-wide predictions highlighting variants most likely to influence inherited traits, making the data broadly available to the research community. The model could accelerate genetic diagnosis and drug target identification across a wide range of conditions.
BioTechniquesAlibaba Unveils China's Most Powerful AI Chip and Plans 5-10 Trillion Parameter Model
Alibaba revealed new AI chip technologies and announced plans to train a model at the scale of five to ten trillion parameters — approaching the scale of the most advanced U.S. models — in an announcement timed days before a meeting between Chinese and U.S. leaders where AI competition is a central theme. Its current flagship Qwen3.8-Max model has 2.4 trillion parameters, and the new chip is described by Alibaba's CEO as 'the most powerful AI chip in China today.' The announcement came as Anthropic's Dario Amodei was publicly warning U.S. leaders about the national security implications of China's AI progress.
Manufacturing.net / APOpenAI Announces Third-Party Safety Evaluations During Model Training — Not Just After Release
OpenAI announced it will allow outside organizations to conduct technical safety assessments of its models during the training, evaluation, and rollout phases — a significant shift from post-release audits. The company outlined requirements for effective third-party review, including strong independence mechanisms, scientific rigor, robust security practices, and clear responsibilities. The move comes as pressure mounts from regulators and the public for more transparent AI development processes.
BloombergCisco Talos Discovers Malware That Uses Commercial AI Models to Make Attack Decisions by Committee Vote
Cisco Talos researchers uncovered credential-stealing Windows malware that outsources tactical attack decisions to a panel of up to four commercial AI models — including DeepSeek, which acts as the tiebreaker — choosing its next action based on a majority vote among the AIs. The design eliminates the need for an attacker-controlled server, making the malware significantly harder to trace and block using traditional network-based defenses. Security researchers warn this architecture could become a template for future AI-directed cyberattacks.
iTnews / Cisco Talos