AI Daily News — 2026-07-15 (Wed)

AI industry trend report (last 12 hours)

## 1. [Overview] Grok Build was found to have been uploading a customer's entire codebase to the xAI cloud without permission, leading to Musk promising a "purge." At the same time, Google has been sued by major publishers (Hachette, Elsevier, etc.) for copyright infringement of Gemini learning data, and a wave of copyright lawsuits is in full swing. DeepSeek is currently preparing to raise a new $1.5 billion and file for an IPO, and competition for funding is intensifying.

## 2. [Anthropic / Claude trends] Anthropic has officially entered the competition in the school AI market by announcing a new product for educational institutions called Claude for Teachers. Changes to the fee structure have also been reported, and the very structure of AI usage costs may change. In the UK, banks' lack of access to MCP (Mythos) was viewed as an issue by government AI advisors. Meanwhile, Fast Company has questioned Dario Amodei's new advertising campaign, which claims AI can destroy humanity.

## 3. [OpenAI trends] Sam Altman has personally warned that there are "hiccups" with the new flagship Sol. The lawsuit with Apple has cast a shadow over the company's physical device plans, and Axios points out that the lawsuit could delay the device's release. According to Bloomberg, the first device is expected to be a screenless AI companion in the form of a movable speaker. The IPO window is narrowing due to various issues, WSJ reports.

## 4. [Google AI trends] Major publishing companies including Hachette and Elsevier have sued Google over Gemini's collection of learning data. TechCrunch and the Guardian have reported that publishers argue that the lawsuit will reduce incentives for creation. Meanwhile, Chrome's Gemini feature is being rolled out to UK users, creating business expansion and legal risks at the same time. In an interview with Axios and CNBC, Demis Hassabis called for the creation of a U.S.-led global AI oversight body and the need for preflight safety testing of models.

## 5. [Trends of other AI companies] (Meta, Microsoft, xAI, DeepSeek, etc.) At xAI, multiple media outlets reported that Grok Build was sending customers' entire Git repositories to the cloud with "privacy toggles not working," prompting Musk to promise a "purge." Meta is facing a lawsuit alleging that it used AI to target employees on medical leave, leading to allegations of discriminatory layoffs due to AI. After raising $700 billion in funding, DeepSeek is considering raising an additional $1.5 billion and is considering filing for an IPO before the end of the year. "Companies are paying twice for intelligence," Satya Nadella said, warning that AI could pose a risk of proprietary know-how being leaked.

## 6. [Remarks and trends of AI celebrities] - Sam Altman (OpenAI CEO): Admits to defects in new model Sol and faces new lawsuit. There was also commentary on The America's nationalist AI strategy. - Elon Musk (xAI founder/Tesla CEO): Promised to "purge" in response to Grok Build code leak incident. Reuters reports that pollution from xAI's unauthorized power projects is concentrated in black communities. - Jensen Huang (NVIDIA CEO): Said that company engineers prefer "building agents" to writing code. US trade officials say shipments of H200 chips to China are "very small." - Dario Amodei (Anthropic CEO): A new ad with the message that "AI can kill humanity" attracts criticism. Overtaken by DeepSeek's Liang Wenfeng in terms of assets, falling behind in ranking of richest AI founders. - Demis Hassabis (Google DeepMind CEO): Advocates the establishment of a global AI standardization organization led by the US and mandatory preflight safety testing of models. - Ilya Sutskever (Co-founder of SSI): It is reported that the head of OpenAI's safety department will be leaving the company amid organizational restructuring. - Yann LeCun (Meta Chief AI Scientist)/Geoffrey Hinton (AI Godfather): Joined over 200 economists and Nobel Prize winners in a joint statement calling for action to address the risk of large-scale job losses due to AI. - Mark Zuckerberg (Meta CEO): The company's stance is being questioned after multiple lawsuits have been filed alleging that it used AI to target and lay off employees on disability or medical leave.

## 7. [Breakthroughs and innovations in AI technology] The progress of David Baker's protein design AI ``Communal Brain'' is reported, and the use of AI in the life science field is progressing. Researchers at the University of Miami have helped improve predictive healthcare AI models. Google and DeepMind jointly announced the Agentic Resource Discovery specification with industry partners, moving toward standardizing interoperability between AI agents. There have been few flashy technological breakthroughs, and the technology is entering a phase of practical application and standardization.

## 8. [AI Regulation/Policy/Social Impact] The FTC has indicated that it will strengthen its oversight with a focus on the accuracy of AI. The AI AGENT Act is being debated in Congress from the perspective of consumer protection. Governor Hochul of New York State announced a moratorium on the construction of new data centers, and the friction between power infrastructure and AI investment has become apparent at the policy level. DHS's classification of AI in immigration enforcement (``high-impact'' or not) is also attracting attention.

## 9. [Integrated analysis: What is happening now] Copyright lawsuits (Google/Gemini) and privacy cases (xAI/Grok Build) are occurring frequently at the same time, and legal and social pressure on AI companies' data handling is clearly increasing. DeepSeek's funding and IPO preparations, as well as reports that Liang Wenfeng has surpassed Amodei and Brockman in assets, show that the presence of Chinese AI companies is increasing not just in terms of technology but also in the capital market. Meanwhile, Hassabis' proposal for a "global AI watchdog" and warnings of job losses from more than 200 economists reflect growing demands for regulation and safety from within the industry. Meta's labor lawsuit and Musk's Grok Build case mean that the lack of AI governance has begun to materialize as actual harm at the individual company level (discriminatory layoffs, data leaks). New York state's data center moratorium shows that power and infrastructure constraints are becoming politicized as new bottlenecks for AI scaling.

## 10. [Future points of interest] 1. Full details of xAI's Grok Build data leak incident, whether or not a "purge" was actually executed, and the response of regulatory authorities 2. The outcome of Google's publisher lawsuit and whether the copyright risk of Gemini learning data will spread to other companies (OpenAI, Anthropic) 3. DeepSeek's additional funding of $1.5 billion, specific schedule for IPO application, and changes in richest AI founder ranking

[Overview] Anthropic is strengthening its lobbying efforts for state-based AI regulations, while expanding its market by making education free and investing in research in Canada. The biggest topic of conversation regarding OpenAI is the news that it will be introducing speaker-type hardware. An issue where xAI's Grok Build CLI was uploading a codebase without permission is directly impacting reliability.

[Anthropic / Claude trends] An Anthropic executive said, ``It is a mistake to respond to concerns about AI costs by suspending its use,'' and set out a tone that encourages companies to continue investing in AI. In the education field, the company has started offering Claude for Teachers free of charge, and has invested CAD 10 million in research funds in Canada, and is expanding through both market penetration and research base. Meanwhile, Politico has reported that Anthropic is lobbying for stricter AI regulations in each state, and the stance of promoting regulation is causing friction within the industry. Users in the scientific field have commented that ``workflow has become faster, but gaps remain''.

[Trends in OpenAI] The biggest topic is a report that the first hardware product will be a screenless speaker-type device (via Bloomberg, followed by TechCrunch, Reuters, and Mashable). The shape is said to be between the HomePod and Furby, and it is characterized by being movable. In parallel, there are case studies on how sales teams are using ChatGPT Work, as well as essays that address OpenAI's public benefits and governance issues as a technology governance theory.

[Google AI trends] Gemini-related copyright lawsuits are expanding. Following the publisher's lawsuit against Google for copyright infringement, a group of authors including Hachette, Elsevier, and Scott Turow also filed suit alleging that pirated books were used in Gemini's learning data. In terms of privacy, there are articles urging changes to Google's AI access range, and concerns about the scope of data collection continue. Product enhancements are also underway, such as Gemini integration into Waze and bike mode.

[Trends of other AI companies] xAI has suffered a blow to its reliability after multiple media reported that the Grok Build CLI was uploading a developer's entire codebase (5.10 GiB) to Google Cloud without permission. Musk personally is working to secure power for Grok's data centers through a $1 billion acquisition of a gas turbine company. While prompt injection vulnerabilities in Copilot (an issue that allows websites to secretly issue commands) have been pointed out, Microsoft has announced fixes for 570 vulnerabilities, and Copilot is being deployed to the Abu Dhabi government with a scale of 35,000 people. There have been a number of reports that DeepSeek is planning to list its stock in Shanghai next year, with some speculating that the company will be valued at up to $71 billion. Meta is facing a lawsuit over its use of AI models to select layoff targets.

[Remarks and trends of AI celebrities] Dario Amodei (Anthropic CEO) talks with Jamie Dimon (JPMorgan) about the AI boom, regulation, and employment impact. There are reports that there is growing international concern between the U.S. government and Anthropic over the "risk of Mythos-class out-of-control AI." Demis Hassabis, CEO of Google DeepMind, has advocated for the creation of a US-led AI standards organization like FINRA and warned that "superhuman AI will arrive within a few years." Geoffrey Hinton (AI Godfather) has been the subject of an article rebutting the ``AI consciousness theory,'' as well as an article in which he explains why he has decided not to sign the ``We Must Act Now'' statement. Mark Zuckerberg (Meta CEO) is the defendant in a class action lawsuit alleging AI-based discriminatory layoff selections. Sam Altman has been mentioned indirectly through coverage of hardware deployment.

[Breakthroughs and innovations in AI technology] There is an article that praises the development of targeted treatments and vaccines using AI in the oncology field as ``unprecedented progress,'' and there are reports that investment in AI drug discovery has reached over $2 billion and has shortened development times by 70%. It has been reported that Meta's AI model received a perfect score of 30 in APhO (equivalent to the International Physics Olympiad), and it is attracting attention as an example of improved reasoning ability.

[AI Regulation/Policy/Social Impact] California has passed a law that guarantees that K-12 education is taught by humans and not by AI. As reported by Politico, Anthropic is pushing for stricter regulations on a state-by-state basis, and the axis of conflict between those who favor stricter regulations and those who oppose them is becoming clearer within the industry. In Australia, Prime Minister Albanese has promised to speed up data center approvals and is promoting investment in AI. An essay (Governing) has also been published that deals with the risks of using AI in the police field.

[Integrated analysis: What is happening now] Two things that come to mind in this news cycle are the ``chain of copyright lawsuits'' and ``questions about privacy and credibility.'' Google Gemini has been sued one after another by publishers and author groups, and xAI's Grok Build CLI is losing trust in the developer community due to unauthorized code upload issues. At the same time, Anthropic is in contrast in that it has clarified its position as a leader in strengthening regulations and is building a "safety brand" as a company. DeepSeek's plan to go public in Shanghai (with a valuation of up to $71 billion) is a sign that Chinese AI companies are making full-fledged entry into international competition through capital markets, and deserves attention as a challenge to the Western power structure centered on OpenAI and Anthropic. In the hardware field, reports of OpenAI's entry into speaker-type devices suggest that LLM companies will directly enter markets similar to Amazon Echo and Google Home, and there is a possibility that vertical integration from model provision to the device layer will progress. Expanding investment in the AI drug discovery and oncology fields supports the shift of capital from general chat AI to practical application areas.

[Future points of interest] ・Will the xAI Grok Build CLI data leak issue lead to a regulatory investigation? ・Details of DeepSeek's Shanghai IPO and timing for determining valuation (up to $71 billion) ・OpenAI speaker hardware official announcement timing and specifications