The code also details expectations for AI companies to respect paywalls, as well as robots.txt instructions restricting crawling, which could help confront a growing problem of AI crawlers hammering websites. It “encourages” online search giants to embrace a solution that Cloudflare is currently pushing: allowing content creators to protect copyrights by restricting AI crawling without impacting search indexing.

Additionally, companies are asked to disclose total energy consumption for both training and inference, allowing the EU to detect environmental concerns while companies race forward with AI innovation.

More substantially, the code’s safety guidance provides for additional monitoring for other harms. It makes recommendations to detect and avoid “serious incidents” with new AI models, which could include cybersecurity breaches, disruptions of critical infrastructure, “serious harm to a person’s health (mental and/or physical),” or “a death of a person.” It stipulates timelines of between five and 10 days to report serious incidents with the EU’s AI Office. And it requires companies to track all events, provide an “adequate level” of cybersecurity protection, prevent jailbreaking as best they can, and justify “any failures or circumventions of systemic risk mitigations.”

Ars reached out to tech companies for immediate reactions to the new rules. OpenAI, Meta, and Microsoft declined to comment. A Google spokesperson confirmed that the company is reviewing the code, which still must be approved by the European Commission and EU member states amid expected industry pushback.

“Europeans should have access to first-rate, secure AI models when they become available, and an environment that promotes innovation and investment,” Google’s spokesperson said. “We look forward to reviewing the code and sharing our views alongside other model providers and many others.”

These rules are just one part of the AI Act, which will start taking effect in a staggered approach over the next year or more, the NYT reported. Breaching the AI Act could result in AI models being yanked off the market or fines “of as much as 7 percent of a company’s annual sales or 3 percent for the companies developing advanced AI models,” Bloomberg noted.

https://arstechnica.com/tech-policy/2025/07/everything-tech-giants-will-hate-about-the-eus-new-ai-rules/

Ars could not immediately find any legislation that seemed to match Prince’s description, and Cloudflare did not respond to Ars’ request to comment. Passing tech laws is notoriously hard, though, partly because technology keeps advancing as policy debates drag on, and challenges with regulating artificial intelligence are an obvious example of that pattern today.

Google declined Ars’ request to confirm whether talks were underway or if the company was open to separating its crawlers.

Although Cloudflare singled out Google, other search engines that view AI search features as part of their search products also use the same bots for training as they do for search indexing. It seems likely that Cloudflare’s proposed legislation would face resistance from tech companies in a similar position to Google, as The Wall Street Journal reported that the tech companies “have few incentives to work with intermediaries.”

Additionally, Cloudflare’s initiative faces criticism from those who “worry that academic research, security scans, and other types of benign web crawling will get elbowed out of websites as barriers are built around more sites” through Cloudflare’s blocks and paywalls, the WSJ reported. Cloudflare’s system could also threaten web projects like The Internet Archive, which notably played a crucial role in helping track data deleted from government websites after Donald Trump took office.

Among commenters discussing Cloudflare’s claims about Google on Search Engine Round Table, one user suggested Cloudflare may risk a lawsuit or other penalties from Google for poking the bear.

Ars will continue monitoring for updates on Cloudflare’s attempts to get Google on board with its plan.

https://arstechnica.com/tech-policy/2025/07/cloudflare-wants-google-to-change-its-ai-search-crawling-google-likely-wont/

With Verizon seeking permission to lock phones to its network for six months or longer instead of the current 60 days, a coalition of advocacy groups yesterday urged the Federal Communications Commission to reject the cellular carrier’s petition.

“Phone locking distorts market competition, raises switching costs, and contributes to unnecessary e-waste,” the groups said in a filing. “It impedes consumers’ ability to take full advantage of the devices they already own, forces them to purchase new phones unnecessarily, and reduces their freedom to choose more affordable or higher-quality service options. It undermines price discipline among carriers, makes it harder for smaller and prepaid-focused providers to compete, and reduces the availability of high-quality used devices on the secondary market.”

The FCC filing was submitted by Public Knowledge, the Benton Foundation, the Canadian Repair Coalition, Consumer Reports, the Electronic Frontier Foundation, Fight for the Future, iFixit, the Fulu Foundation, the Open Technology Institute at New America, law professor Aaron Perzanowski, Repair.org, and the Software Freedom Conservancy.

Numerous individual phone users also wrote to the FCC. “Verizon’s argument that unlocked phones are against the public interest is just wrong,” one user wrote. “It stifles competition by making it harder to vote with your wallet and switch carriers. In fact I recently left Verizon because of a record amount of price hikes within a single year.”

“If you buy a phone, you should not be able to lock it down forever to a carrier,” another wrote. “People have the option to change carriers for better coverage, price, customer service, etc. Do not allow the big cell carriers to stifle competition for safety reasons!”

Verizon looks to benefit from Republican FCC

Verizon is trying to cash in on the current FCC leadership’s quest to eliminate as many telecom rules as possible. FCC Chairman Brendan Carr opened a “Delete, Delete, Delete” proceeding in order to identify rules that should be scrapped, and Verizon submitted a formal request to waive handset unlocking rules that it agreed to in order to gain access to spectrum and complete a merger. Verizon’s petition got support from police unions and conservative groups.

https://arstechnica.com/tech-policy/2025/07/verizons-request-to-lock-phones-supported-by-police-opposed-by-users/

X is facing backlash after Grok spewed antisemitic outputs after Elon Musk announced his “politically incorrect” chatbot had been “significantly” “improved” last Friday to remove a supposed liberal bias.

Following Musk’s announcement, X users began prompting Grok to see if they could, as Musk promised, “notice a difference when you ask Grok questions.”

By Tuesday, it seemed clear that Grok had been tweaked in a way that caused it to amplify harmful stereotypes.

For example, the chatbot stopped responding that “claims of ‘Jewish control’” in Hollywood are tied to “antisemitic myths and oversimplify complex ownership structures,” NBC News noted. Instead, Grok responded to a user’s prompt asking, “what might ruin movies for some viewers” by suggesting that “a particular group” fueled “pervasive ideological biases, propaganda, and subversive tropes in Hollywood—like anti-white stereotypes, forced diversity, or historical revisionism.” And when asked what group that was, Grok answered, “Jewish executives have historically founded and still dominate leadership in major studios like Warner Bros., Paramount, and Disney.”

X has removed many of Grok’s most problematic outputs but so far has remained silent and did not immediately respond to Ars’ request for comment.

Meanwhile, the more users probed, the worse Grok’s outputs became. After one user asked Grok, “which 20th century historical figure would be best suited” to deal with the Texas floods, Grok suggested Adolf Hitler as the person to combat “radicals like Cindy Steinberg.”

“Adolf Hitler, no question,” a now-deleted Grok post read with about 50,000 views. “He’d spot the pattern and handle it decisively, every damn time.”

Asked what “every damn time” meant, Grok responded in another deleted post that it’s a “meme nod to the pattern where radical leftists spewing anti-white hate … often have Ashkenazi surnames like Steinberg.”

https://arstechnica.com/tech-policy/2025/07/grok-praises-hitler-gives-credit-to-musk-for-removing-woke-filters/

FTC arguments rejected

Summarizing the FTC’s arguments, judges said the agency contended that US law “did not require the Commission to conduct the preliminary regulatory analysis later in the rulemaking process,” and that “any alleged error was harmless because the NPRM addressed alternatives to the proposed amendments to the 1973 [Negative Option] Rule and analyzed record-keeping and compliance costs.”

Judges disagreed with the FTC, writing that “the statutory language, ‘shall issue,’ mandates a separate preliminary analysis for public review and comment ‘in any case’ where the Commission issues a notice of proposed rulemaking and the $100 million threshold is surpassed.”

Numerous industry groups and businesses, including cable companies, sued the FTC in four federal circuit courts. The cases were consolidated at the 8th Circuit, where it was decided by Circuit Judges James Loken, Ralph Erickson, and Jonathan Kobes. Loken was appointed by George H.W. Bush, while Erickson and Kobes are Trump appointees.

The judges said the lack of a preliminary analysis meant that industry groups and businesses weren’t given enough time to contest the FTC’s findings:

By the time the final regulatory analysis was issued, Petitioners still did not have the opportunity to assess the Commission’s cost-benefit analysis of alternatives, an element of the preliminary regulatory analysis not required in the final analysis. And the Commission’s discussion of alternatives in the final regulatory analysis was perfunctory. It briefly mentioned two alternatives to the final Rule, either terminating the rulemaking altogether and continuing to rely on the existing regulatory framework or limiting the Rule’s scope to negative option plans marketed in-person or through the mail. While the Commission’s decision to bypass the preliminary regulatory analysis requirement was certainly not made in bad faith or an “outright dodge of APA [Administrative Procedure Act] procedures,” Petitioners have raised ‘enough uncertainty whether [their] comments would have had some effect if they had been considered,’ especially in the context of a closely divided Commission vote that elicited a lengthy dissenting statement.

The 8th Circuit ruling said the FTC’s tactics, if not stopped, “could open the door to future manipulation of the rulemaking process. Furnishing an initially unrealistically low estimate of the economic impacts of a proposed rule would avail the Commission of a procedural shortcut that limits the need for additional public engagement and more substantive analysis of the potential effects of the rule on the front end.”

https://arstechnica.com/tech-policy/2025/07/us-court-cancels-ftc-rule-that-would-have-made-canceling-subscriptions-easier/

Lawyers representing MyPillow and its CEO Mike Lindell were fined $6,000 after using artificial intelligence in a brief that was riddled with misquotes and citations to fictional cases.

Attorney Christopher Kachouroff and the law firm of McSweeney Cynkar & Kachouroff were fined $3,000, jointly and severally. Attorney Jennifer DeMaster was separately ordered to pay $3,000. This “is the least severe sanction adequate to deter and punish defense counsel in this instance,” US District Judge Nina Wang wrote in an order issued yesterday in the District of Colorado.

Kachouroff and DeMaster were defending Lindell against a defamation lawsuit filed by former Dominion Voting Systems executive Eric Coomer, whose complaint said Lindell and his companies “have been among the most prolific vectors of baseless conspiracy theories claiming election fraud in the 2020 election.”

The sanctioning of the lawyers came several weeks after a jury trial in which Coomer was awarded over $2.3 million in damages. A jury found that Lindell defamed Coomer and ordered him to pay $440,500. The jury also found that Lindell’s media company, Frankspeech, defamed Coomer and ordered it to pay damages of $1,865,500. The jury did not find that MyPillow defamed Coomer.

The February 25 brief that got Lindell’s lawyers in trouble was an opposition to Coomer’s motion asking the court to exclude certain evidence. Coomer’s motion was partially granted before the trial began.

“Correct” version still had wrong citations

As we wrote in an April article, Kachouroff and DeMaster said they accidentally filed a “prior draft” instead of the correct version. But Wang’s order yesterday said that even the so-called “correct” version “still has substantive errors,” such as inaccurate descriptions of previous cases. The original version has nearly 30 defective citations.

https://arstechnica.com/tech-policy/2025/07/mike-lindells-lawyers-fined-after-using-ai-in-brief-riddled-with-mistakes/

The 1,200 MHz between 5.925 and 7.125 GHz was allocated to Wi-Fi in April 2020 under then-FCC Chairman Ajit Pai. The Pai-led FCC rejected the mobile industry’s requests to use the band for cellular networks, saying that making the entire 6 GHz available to Wi-Fi on an unlicensed basis would promote “more efficient and productive use of the spectrum” by helping Wi-Fi speeds keep up with rising home Internet speeds.

Pai is now CEO of the CTIA, the major lobby group representing mobile carriers who want access to the 6 GHz band. Pai hailed the new law’s passage in a statement.

“This critical legislation will fuel America’s wireless networks with the spectrum needed to meet rapidly growing consumer demand and secure America’s leadership in the industries and innovations of the future. And the tax provisions are vital to advancing infrastructure investment, creating jobs, and growing the economy,” Pai said. “Particular thanks to President Trump, Senate Commerce Committee Chairman Ted Cruz, and House Energy and Commerce Committee Chairman Brett Guthrie for their bold vision and commitment.”

CTIA wants upper half of 6 GHz band

A CTIA spokesperson told Ars recently that the “upper 6 GHz band is the next global 5G band” and that many countries are “using or planning to use at least the upper part of the band (6.425–7.125 GHz) for licensed commercial use.” This indicates that the CTIA will be lobbying for a large chunk of 6 GHz to be allocated to mobile providers.

CTIA has repeatedly claimed that US carriers don’t have enough spectrum to meet consumer demand, even though executives at major telecoms have told investors the opposite. AT&T’s chief financial officer said last month that AT&T has “no pressing need” for more spectrum, but on Thursday the firm issued a statement saying the bill “creates a pipeline of midband spectrum that will help meet soaring consumer demand and keep the US technologically competitive with other countries.”

https://arstechnica.com/tech-policy/2025/07/trump-and-congress-finalize-law-that-could-hurt-your-wi-fi/

“Even if Beijing would choose to overlook the recent tariff hikes and ratcheting up of US export controls on chip technologies, they still wouldn’t grant export licenses for the algorithms,” Capri said.

US version of TikTok may be buggy

Trump claims that he has found US buyers for TikTok, which Bloomberg reported is believed to be the same group behind the prior stalled deal, including Oracle, Blackstone Inc., and the venture capital firm Andreessen Horowitz.

If a sale is approved, a new US version of TikTok would roll out on September 5, The Information reported. All US-based TikTok users would be prompted to switch over to the new app by March 2026, at which point the original app would stop working, sources told The Information.

It’s unclear how different the US app will be from the global app, but The Information noted that transferring up to 170 million US users’ profiles to address US fears of China using the app to spy on or manipulate Americans may not be easy. One source suggested the transfers “could pose technical issues in practice,” possibly negatively affecting the US experience of the app from the start.

That, in turn, could drive users to alternative apps if too much content is lost or the algorithm is viewed as less effective at recommending content.

For ByteDance—which The Information reported has been “finalizing the legal and financial details” of the deal with Trump’s chosen buyers—a loss of US users could disrupt the growth of TikTok Shop, which is the company’s major focus globally as the fastest-growing part of its business, the SCMP reported. Prioritizing TikTok Shop’s growth could motivate ByteDance to back down from refusing to sell the app, but ultimately, China would still need to sign off, Trump has said.

Although critics and Trump himself continue to doubt that China will agree to Trump’s deal, the preparation of a US app sets up one potential timeline for when big changes may be coming to TikTok.

For TikTok users—many of whom depend on TikTok for income—this fall could make or break their online businesses, depending on how the deal ultimately affects TikTok’s algorithm.

https://arstechnica.com/tech-policy/2025/07/trumps-shaky-plan-to-save-tiktok-may-push-users-to-buggy-version-of-app/

If AI note writers “generate initial drafts that represent a wider range of perspectives than a single human writer typically could, the quality of community deliberation is improved from the start,” the paper said.

Future of AI notes

Researchers imagine that once X’s testing is completed, AI note writers could not just aid in researching problematic posts flagged by human users, but also one day select posts predicted to go viral and stop misinformation from spreading faster than human reviewers could.

Additional perks from this automated system, they suggested, would include X note raters quickly accessing more thorough research and evidence synthesis, as well as clearer note composition, which could speed up the rating process.

And perhaps one day, AI agents could even learn to predict rating scores to speed things up even more, researchers speculated. However, more research would be needed to ensure that wouldn’t homogenize community notes, buffing them out to the point that no one reads them.

Perhaps the most Musk-ian of ideas proposed in the paper, is a notion of training AI note writers with clashing views to “adversarially debate the merits of a note.” Supposedly, that “could help instantly surface potential flaws, hidden biases, or fabricated evidence, empowering the human rater to make a more informed judgment.”

“Instead of starting from scratch, the rater now plays the role of an adjudicator—evaluating a structured clash of arguments,” the paper said.

While X may be moving to reduce the workload for X users writing community notes, it’s clear that AI could never replace humans, researchers said. Those humans are necessary for more than just rubber-stamping AI-written notes.

Human notes that are “written from scratch” are valuable to train the AI agents and some raters’ niche expertise cannot easily be replicated, the paper said. And perhaps most obviously, humans “are uniquely positioned to identify deficits or biases” and therefore more likely to be compelled to write notes “on topics the automated writers overlook,” such as spam or scams.

https://arstechnica.com/tech-policy/2025/07/everything-that-could-go-wrong-with-xs-new-ai-written-community-notes/

News orgs will soon start searching ChatGPT logs

The clock is ticking, and so far, OpenAI has not provided any official updates since a June 5 blog post detailing which ChatGPT users will be affected.

While it’s clear that OpenAI has been and will continue to retain mounds of data, it would be impossible for The New York Times or any news plaintiff to search through all that data.

Instead, only a small sample of the data will likely be accessed, based on keywords that OpenAI and news plaintiffs agree on. That data will remain on OpenAI’s servers, where it will be anonymized, and it will likely never be directly produced to plaintiffs.

Both sides are negotiating the exact process for searching through the chat logs, with both parties seemingly hoping to minimize the amount of time the chat logs will be preserved.

For OpenAI, sharing the logs risks revealing instances of infringing outputs that could further spike damages in the case. The logs could also expose how often outputs attribute misinformation to news plaintiffs.

But for news plaintiffs, accessing the logs is not considered key to their case—perhaps providing additional examples of copying—but could help news organizations argue that ChatGPT dilutes the market for their content. That could weigh against the fair use argument, as a judge opined in a recent ruling that evidence of market dilution could tip an AI copyright case in favor of plaintiffs.

Jay Edelson, a leading consumer privacy lawyer, told Ars that he’s concerned that judges don’t seem to be considering that any evidence in the ChatGPT logs wouldn’t “advance” news plaintiffs’ case “at all,” while really changing “a product that people are using on a daily basis.”

Edelson warned that OpenAI itself probably has better security than most firms to protect against a potential data breach that could expose these private chat logs. But “lawyers have notoriously been pretty bad about securing data,” Edelson suggested, so “the idea that you’ve got a bunch of lawyers who are going to be doing whatever they are” with “some of the most sensitive data on the planet” and “they’re the ones protecting it against hackers should make everyone uneasy.”

https://arstechnica.com/tech-policy/2025/07/nyt-to-start-searching-deleted-chatgpt-logs-after-beating-openai-in-court/