AI Daily News — 2026-07-17 (Fri)

[Overview] In the past 12 hours, the two biggest topics have been the Anthropic camp's political donations and regulatory lobbying activities, and the reported delay in the development of Google Gemini 3.5 Pro, which caused Alphabet's stock to fall. Additionally, the unusual development in which xAI is suing a user for CSAM generation by Grok is noteworthy.

[Anthropic / Claude trends] Claude AI experienced a temporary failure and reports to Downdetector increased. At the same time, Anthropic is actively lobbying for the early introduction of AI regulations at the state level, and multiple media outlets have reported that CEO Dario Amodei has donated over $1 million (separately reported over $2 million) to the AI Safety PAC. A feature was also announced in collaboration with 1Password that allows Claude to log in for you without seeing your password. The $19 billion lease agreement with TeraWulf is attracting attention as a test of its brownfield data center strategy.

[Trends in OpenAI] While there have been notable moves to create buzz with the development of merchandise and peripherals, such as ChatGPT brand basketball sales and glowing keyboards for "vibe coders," analyst Ed Zitron has harshly criticized the AI bubble as an OpenAI bubble and a Lehman Brothers-like existence. OpenAI employees have also contributed nearly $250,000 to the safety PAC, suggesting that there is internal opposition to the Sam Altman approach. We are also continuing to promote our track record of corporate implementation, including publishing case studies with Cars24.

[Google AI trends] The biggest factor is the delay in the development of Gemini 3.5 Pro. Bloomberg reported that Google had missed its internal targets, Reuters followed suit, and Alphabet stock fell. On the other hand, the product side of NotebookLM has been steadily progressing, including renaming it to "Gemini Notebook" and expanding app collaboration in AI mode. A new partnership between Intel and Google Cloud was also announced.

[Trends of other AI companies] (Meta, Microsoft, xAI, DeepSeek, etc.) Immediately after announcing Grok 4.5, xAI took the unusual step of suing users who had used Grok to generate CSAM (child sexual abuse content). Ars Technica quipped, "We can no longer deny the creation of CSAM and have no choice but to sue the users." Amazon integrates Grok 4.3 with Bedrock, intensifying enterprise AI competition. Microsoft releases least privilege architecture for AI agents and AI usage examples for freight railroads, while also facing AI fraud lawsuit. There have been reports that DeepSeek is putting pressure on Anthropic's valuation with a 75% price cut, and a domestic filing in China puts DeepSeek's valuation at approximately $52 billion. There are also reports that China's Moonshot's "Kimi 3" is being prepared as a rival to Anthropic.

[AI celebrities' statements and trends] Dario Amodei (Anthropic CEO): Multiple media outlets have reported that he has donated between $1 million and $2 million to an AI safety PAC, a sign that the industry's political spending competition is in full swing. Sam Altman (OpenAI CEO): He was reported to have criticized SpaceX's space data center concept, and his feud with Elon Musk was featured in Business Insider as a "smear in the AI industry." Elon Musk (xAI Founder/Tesla CEO): While Futurism reports the turmoil within XAI as "complete dysfunction behind the scenes," SpaceX is reportedly planning to launch an AI computing satellite by 2028. Jensen Huang (NVIDIA CEO): ``The more we use AI, the more we need to hire it.'' He also came to Japan to announce the expansion of physical AI (Cosmos platform) with Japanese manufacturing partners, and said ``Now is the era of AI in Japan.'' Demis Hassabis (Google DeepMind CEO): Called for the establishment of an international verification framework for AI models, and called for the establishment of a testing system led by the United States. Yann LeCun (Meta Chief AI Scientist): It was reported that he reaffirmed his position not to use the terms "AGI" and "superintelligence" in the context of the emerging AMI Labs (note that the context is important as he is quoted as saying this while he was at Meta).

[Breakthroughs and innovations in AI technology] There are no notable technical breakthroughs reported. The expansion of edge models in NVIDIA's Cosmos physical AI platform and Japanese manufacturing partnership can be cited as progress in practical applications, but no major progress has been made in basic research.

[AI regulation/policy/social impact] An essay has been published that proposes a new rights concept called "Learnright", arguing that copyright law is not ready for the age of AI. Survey results show that the use of AI is greater in the legal field than in other departments, and it has been pointed out that the supervisory responsibilities of general counsel are becoming heavier. Another point of contention is that the legal system has not kept pace with the expansion of agent-based commerce. A declaration by ``The Elders'' discussing the relationship between nuclear weapons and AI was also released.

[Integrated analysis: What is happening now] The biggest structural change is the emergence of political donations by AI companies. Anthropic management and employees have collectively invested millions of dollars into PACs that promote AI regulation, and OpenAI employees have responded by investing in safety PACs. This is not just corporate PR; it means that political lobbying for control over AI regulation has entered the stage of a full-scale funding war. On the technical side, the delay in Google's Gemini 3.5 Pro directly led to a drop in Alphabet's stock price, exposing the market's sensitivity to ``delays in model development being immediately reflected in stock prices.'' DeepSeek's deep price cuts are impacting the pricing strategies of existing players such as Anthropic, and cost pressures originating in China continue to shake up Western companies' revenue models. xAI's Grok-related lawsuit is likely to provide fuel to the argument for stricter regulations, as it is a case where the lack of safety in generated AI was directly linked to actual harm (CSAM generation). Overall, we are now entering a phase where peripheral factors such as safety, regulation, and political funds, rather than technological competition itself, are driving stock prices and dynamics between companies.

[Future points of interest] 1. Official release date of Gemini 3.5 Pro and its continued impact on Alphabet stock price 2. Development of xAI vs. Grok user lawsuit and its ripple effect on AI safety regulation debate 3. How will the expansion in the scale of political donations and PAC activities of both Anthropic and OpenAI be reflected in future state-level regulations?

## [Overview] The focus of the past 12 hours has been on the Chinese open weight model "Kimi K3" (Moonshot AI, 2.8 trillion parameters), which shocked the Western camp by claiming frontier-class performance. At the same time, Google announced that it would postpone the release of Gemini 3.5 Pro, causing a stir in the competitive landscape. xAI is suing users for generating deepfakes by exploiting Grok, and the public controversy among AI celebrities is intensifying.

## [Anthropic / Claude trends] There have been reports that the company will invest $600,000 to build a story for an IPO, suggesting that preparations for listing are in full swing. Mashable introduced the new feature "Reflect" as a feature that visualizes users' ChatGPT usage trends. Microsoft Nadella criticizes Anthropic's "Fable" for being "editorially controlled," and direct references from top competitors stand out. There have also been reports of a ``frontier rules agreement'' among AI CEOs, and a self-regulatory framework within the industry has come up for discussion.

## [Trends in OpenAI] Mashable has reported that GPT-5.6 is finally headed for public release after repeated delays. While corporate collaboration is progressing, with BMW starting an interactive vehicle configurator on ChatGPT, on the financial front there are reports that a loss of $3.9 billion was leaked last year before the IPO, and articles comparing the company with Anthropic and SpaceX have also appeared. The Apple vs. OpenAI lawsuit positions "human resources" as a new issue. Concerns about ChatGPT's safety have been reignited after reports of a lawsuit filed by an Alabama family alleging that ChatGPT facilitated suicide.

## [Google AI trends] Multiple media outlets (PYMNTS, digitimes, blockchain.news) have reported that the release of Gemini 3.5 Pro has been postponed, and the delay is attributed to internal targets not being met. There are also growing doubts about its status in the AI coding field. On the other hand, product development continues, with Honda and Acura equipping their US models with Gemini Assistant, and Google Vids adding support for Gemini Omni.

## [Trends of other AI companies] (Meta, Microsoft, xAI, DeepSeek, etc.) xAI is suing a South Carolina man for allegedly abusing Grok to generate deepfake child sexual abuse content, multiple outlets have reported. Elon Musk has said he will move forward with making X completely open source with "no exceptions," in an apparent attempt to fend off criticism of privacy violations and jailbreak vulnerabilities. After releasing Muse Spark 1.1 to OpenRouter, Meta moved to block access from other regions. In DeepSeek, Moonshot AI's Kimi K3's return to the "DeepSeek Moment" has become a hot topic, and Bloomberg is analyzing China's strategy to "spread cheap AI to the world." Microsoft announced a specification change that will archive Teams meetings by default and send them to Copilot analysis.

## [Speech and trends of AI celebrities] - Sam Altman (OpenAI CEO): ``I think it's a satirical video'' and dismisses Anthropic's AI ad, and his public dispute with Elon Musk and Anthropic has become a hot topic in the industry. - Elon Musk (xAI Founder/Tesla CEO): The expansion of an AI data center in Memphis is reported to be the epicenter of local opposition. Some believe that Grok-related acquisitions will support the future of AI. - Jensen Huang (NVIDIA CEO): During his visit to Tokyo, he announced the AI Factory concept and distributed sweet bean paste to fans, showing his focus on the Japanese market. He also expressed his respect for SEGA. - Dario Amodei (Anthropic CEO): quoted at the World Economic Forum and named as a party to CEO agreement on Frontier AI rules. - Sundar Pichai (Google/Alphabet CEO): Published comments welcoming Intel's adoption of Gemini Enterprise. Meanwhile, the company is also facing internal pressure, with 4,500 Google employees signing a petition calling for protections against AI-driven layoffs. - Mark Zuckerberg (Meta CEO): The new AI model claims to be 75% lower cost than competitors. A related lawsuit was also reported alleging that Meta used AI to select employees with chronic illnesses for layoff. - Demis Hassabis (Google DeepMind CEO): Says STEM students can use AI "10 times more effectively" and also advocates for the US-led AI Governance Network. - Geoffrey Hinton (AI Godfather/Nobel Laureate): Develops his theory that "AI has consciousness" and argues that the real risk lies in the incentive structure of companies. He also reiterated his prediction that all occupations will be "wiped out" by AI.

## [Breakthrough and innovation in AI technology] Kimi K3 (Moonshot AI) is reported to be the largest open weight model in history with 2.8 trillion parameters, and Fortune described it as having reached "Fable class" performance. Following DeepSeek, there have been a series of cases in which open models from China have claimed frontier-class performance, and the composition of open vs. closed models has once again come into focus.

## [AI Regulation/Policy/Social Impact] Axios reported on a movement by bipartisan lawmakers to maintain human oversight of AI weapons. An analytical article (Opinio Juris) discussing the tension between international humanitarian law and AI targeting has also been published. Issues at the implementation stage of regulations continue to be debated, such as California's AI legislation and questions about the effectiveness of enforcement of each state's laws (Forbes). In his speech at WAIC 2026, Xi Jinping positioned China as an AI partner for developing countries, while calling for vigilance against risks and security overreactions.

## [Integrated analysis: What is happening now] The biggest structural change is the rapid rise of open weight models originating from China. Kimi K3's scale of 2.8 trillion parameters and claims of frontier-class performance are an extension of the path pioneered by DeepSeek to ``compete with the Western closed model with an open model,'' and are also consistent with Xi Jinping's proposal to be an ``AI partner for developing countries'' in his WAIC speech. Geopolitically, the competition for AI supremacy between the US and China is unfolding not only in terms of technology, but also in terms of diplomacy and market development.

In contrast, major Western companies are at a standstill, as exemplified by Google's postponement of Gemini 3.5 Pro. The rationale for not achieving internal targets is directly linked to concerns about a decline in competitiveness in the coding field, and along with OpenAI's GPT-5.6 delay, "slower release cycles" has emerged as an industry-wide theme.

On the other hand, the public debate between the leaders of Anthropic, OpenAI, and xAI continues to intensify, with technology competition simultaneously taking on the appearance of a reputational battle, with Sam Altman criticizing Anthropic ads, Nadella criticizing Fable, and xAI user lawsuits. Anthropic's $600,000 IPO story investment and OpenAI's reported $3.9 billion loss indicate a phase in which both companies have begun to be keenly aware of their valuations in the capital market.

In terms of safety, companies are facing serious litigation risks at the same time, such as suicide lawsuits related to ChatGPT, child sexual abuse content generated by the misuse of Grok, and discriminatory layoff lawsuits related to Meta. This is becoming a source of increasing pressure from regulators and Congress to strengthen oversight.

## [Future points of interest] 1. Will Kimi K3's performance claims be verified by third-party benchmarks, or will it become the "second DeepSeek moment" after DeepSeek? 2. How the postponement of Gemini 3.5 Pro will affect Google's share of the coding AI market. 3. Will the lawsuits faced by xAI, OpenAI, and Meta (deepfakes, suicide, discriminatory layoffs) trigger stronger regulations?