President Trump signed an executive order on Tuesday establishing a voluntary framework for government review of frontier AI models before public release, ending weeks of internal White House conflict over how aggressively to regulate the technology. The order, titled “Promoting Advanced Artificial Intelligence Innovation and Security,” was signed privately without the usual livestream or public ceremony, a contrast with the fanfare that typically accompanies presidential AI announcements.

The final version is substantially narrower than the draft Trump rejected on 21 May, when he scrapped a planned signing ceremony over concerns that the order “could dull America’s edge on AI technology.” The original draft proposed a 90-day mandatory pre-release review period and would have given the government formal evaluation authority over frontier models. The signed version asks companies to voluntarily submit models 30 days before release and participate in a collaborative framework rather than submitting to mandatory testing.

What the order does

The executive order establishes three main mechanisms. First, a voluntary pre-release review framework in which AI developers can engage the government to determine whether models under development qualify as “covered frontier models,” provide access for up to 30 days before planned release, and collaborate on selecting “trusted partners” for early access. The framework is explicitly voluntary, meaning companies can decline to participate without penalty.

Second, the order creates an AI cybersecurity clearinghouse within 30 days, coordinated by the Treasury Secretary, the National Cyber Director, the NSA, and CISA. The clearinghouse will scan for software vulnerabilities, validate discoveries, and coordinate remediation and patch distribution, a direct response to the Mythos crisis that demonstrated how AI-discovered vulnerabilities can outpace existing disclosure and patching processes.

Third, federal agencies are directed to develop benchmarks for assessing AI models’ cybersecurity capabilities and to strengthen the government’s own security defences against AI-enabled threats. The order also addresses AI safety research, though the specific provisions are less prescriptive than what the original draft contained.

What was cut

The differences between the scrapped draft and the signed order reflect the victory of the pro-industry faction within the White House. The 90-day mandatory review was reduced to a 30-day voluntary window. The formal government evaluation authority was replaced with a collaborative framework. The reporting requirements for companies developing powerful models, which would have echoed provisions in Biden’s repealed AI executive order, were softened to avoid what industry allies characterised as regulatory overreach.

Silicon Valley’s objections to the original draft were decisive. AI companies argued that mandatory pre-release testing would slow American innovation, create a competitive disadvantage relative to Chinese firms facing no equivalent requirements, and establish a precedent for government gatekeeping of technology deployment. The signed order addresses those concerns by making participation voluntary and framing the government’s role as collaborative rather than regulatory.

The gap it leaves

The voluntary framework means the order’s effectiveness depends entirely on whether AI companies choose to participate. Companies already engaged in pre-release testing with CAISI, including Google, Microsoft, and xAI, may continue or expand that cooperation. Companies that view government review as commercially disadvantageous or that are racing to ship products can simply opt out.

The EU’s AI Act, entering full enforcement in August, provides a stark contrast: mandatory requirements, statutory authority, and penalties for non-compliance. The Trump order establishes norms and creates institutional infrastructure (the cybersecurity clearinghouse, the benchmark development process) but relies on goodwill rather than obligation.

For the White House, the quiet signing may be the point. The order gives the administration a policy document it can reference when asked about AI oversight, creates structures that could be strengthened later, and avoids a public confrontation with an AI industry whose leaders are among the administration’s most visible supporters. Whether a voluntary framework is adequate for a technology that can discover 10,000 zero-day vulnerabilities in a month is the question the order deliberately leaves unanswered.

Apple TV‘s hit sci-fi series is back, and the new Silo season 3 trailer makes it clear that the show is finally ready to answer its biggest question: how did the world end?

The 10-episode third season premieres on July 3, with new episodes dropping every Friday through September 4. And if this trailer is anything to go by, the wait has been worth it.

Juliette is back, but she’s not quite herself

Rebecca Ferguson’s Juliette survived the incinerator at the end of season 2, but she didn’t walk away unscathed. She’s lost all her memories, and that’s where things get truly exciting.

Camille Sims (Alexandria Riley) has been feeding her a false narrative for three months, and mysterious pills spotted in the trailer appear to be playing a role in keeping those memories buried. A voice identified only as The Algorithm confirms it, telling Camille that Juliette has had no memories other than the ones she was given.

There’s even a cryptic note warning Juliette not to take the mysterious pills, so someone out there wants her to remember. This memory manipulation is a bold departure from Hugh Howey’s Wool novels, and it adds another layer of tension.

The trailer also suggests that Juliette will become the mayor of the silo in season 3, settling into a powerful role she has no memory of earning. On top of all that, a countdown in the trailer hints that the silo itself is running out of time.

As one character ominously puts it, the amount of panic, if people knew how little time they had left, would be unthinkable. Meanwhile, Steve Zahn also makes a welcome return as Solo, whose fate was left uncertain at the end of season 2.

The past timeline is where Silo season 3 gets really interesting

While Juliette’s arc drives the present-day story, the trailer’s most exciting reveal is its deep dive into the past. Journalist Helen Drew (Jessica Henwick) and Congressman Daniel Keene (Ashley Zukerman), who were briefly introduced in the season 2 finale, are now central characters.

Their storyline is set centuries before the events of the silo and draws heavily from Shift, the second book in Howey’s trilogy. The two uncover a conspiracy involving what appears to be a radiological weapon attack, a chain of events that ultimately forced humanity underground. As the trailer tagline puts it, “the truth lies in the past.”

Joining the cast this season are Laura Innes, Jessica Brown Findlay, Morven Christie, Reed Birney, Matt Craven, and Colin Hanks. Silo season 3 looks like the show has rediscovered everything that made season 1 so gripping – the paranoia, the mystery, and that nagging sense that nobody can be trusted.

With an origin story that goes back centuries now in play, this is shaping up to be its most ambitious chapter yet. If you haven’t started watching, now is the perfect time to catch up before July 3.

from the age-gating-is-still-bad dept

After public outrage, California lawmakers are moving closer to exempting open-source operating systems from the sweeping age-bracketing regime mandated by last year’s Digital Age Assurance Act (AB 1043). Nonetheless, the current bill still jeopardizes internet users’ speech, privacy, and security.

While the open source exemption, if passed, would improve the law, the remaining amendments proposed by AB 1856 would require all web browsers and websites to request and collect users’ ages. This is an expansion of last year’s AB 1043’s age-bracketing system that compounds its constitutional harms to users’ speech, privacy, and security. As AB 1856 moves on to the Senate, EFF will continue fighting for amendments that reduce those harms.

AB 1856 Extends AB 1043’s Age-Gating Regime

Last year, California passed AB 1043, which requires all operating systems and app stores to create age-bracketing systems that segment users based on their ages. As we’ve written, that regime is a recipe for censorship: it creates unnecessary and unconstitutional barriers to accessing lawful online speech, threatens our right to anonymity, and pressures online services to collect troves of valuable and sensitive user data. On top of that, A.B. 1043’s wide-sweeping compliance burdens impose disproportionate harms on the open-source ecosystem that underpins much of the modern web. 

Given these flaws, lawmakers introduced AB 1856 this year as a supposed “clean-up” bill for AB 1043. But instead of sticking to fixing AB 1043’s unique and serious harms (like its impact on open-source operating systems), AB 1856 also expanded the regime even further—extending its age-bracketing requirements beyond operating systems and app stores to browsers and websites. 

EFF opposed AB 1856 on two grounds, which we explained in our opposition letter to the Assembly: 

1. The harms that age-gating regimes pose to users’ speech, privacy, and anonymity; and
2. The disproportionate harms that this particular regime imposes on open-source developers. 

Open Source Concerns Somewhat Alleviated By Amendment

On May 28th, AB 1856 passed the Assembly in a nearly unanimous vote (68-1). 

Before that vote, however, AB 1856 was amended to relieve the compliance burden on open-source operating systems. This is a meaningful improvement and a welcome relief for open-source developers, who have been loud and clear about how much of an existential threat A.B. 1043’s age-gating mandate would pose.

The new exception reads:

“Operating system provider” does not mean a person or entity that distributes an operating system or application under license terms that permit a recipient to copy, redistribute, and modify the software.”

EFF understands this amendment to exempt open-source operating systems from the requirement to collect and transmit users’ age-bracket data. That is a definite win for open-source developers. The bill is narrower now than it was before, and lawmakers clearly responded to concerns raised by EFF and the broader open-source community. 

Some important questions still remain—for example, it is unclear how the law would apply when an open-source operating system is incorporated into a commercial product or service. And, given the structure of where the exemption is placed under the “operating system provider” definition, lawmakers could stand to clarify that the exemption applies to open-source operating systems and applications.

Nonetheless, that ambiguity aside, this amendment does substantially reduce the threat that AB 1043 could have on many open-source developers. 

AB 1856 Still Expands the Problematic Age-Bracketing Regime

Don’t get us wrong—if this bill passes, we will be very happy that AB 1043 does not pose nearly the amount of harm to our friends behind open-source operating systems. But even after these amendments, EFF remains opposed to AB 1856 because it ultimately expands California’s sweeping age-bracketing framework far beyond the original scope of AB 1043. 

In AB 1856 and its amendments, the Assembly failed to address the core problem with AB 1043’s age-bracketing regime: mandated age-gating systems threaten users’ speech, privacy, anonymity, and security. 

Even though AB 1043 does not explicitly require companies to perform age verification, it nonetheless imposes a liability structure that strongly pressures companies to verify users’ ages anyway. In practice, that could lead to more ID checks, more biometric scanning, more invasive data collection and risk of breach, and more barriers to adults’ and young people’s lawful speech.

In fact, instead of narrowing AB 1043’s wide net, AB 1856 expanded it to add browser providers and website operators to the list of entities that must comply with its age-bracketing requirements. This dramatically broadens the scope of AB 1043 and pulls more services, developers, and users into an anonymity- and privacy-destroying data collection framework that has not yet been implemented or evaluated. The result would make it nearly impossible for regular internet users to avoid AB 1043’s age gates.

The Fight Moves to the Senate

On those grounds, EFF will continue to oppose AB 1856. Though it has passed the Assembly, the fight is not over. As the bill moves through the Senate, we’ll continue to push for amendments that actually “clean up” and narrow the scope of AB 1043, and offer more protection to users from the harms of age-gating systems.

Republished from the EFF’s Deeplinks blog.

Filed Under: ab 1043, ab 1856, age verification, california, open source, operating systems, websites

• Rivian’s boss believes we will have eyes-off driving within 18 months
• It will be the “most disruptive feature we’ve seen”, according to RJ Scaringe
• But a new report suggests Tesla engineers and staff don’t trust the technology

Rivian’s boss and CEO, RJ Scaringe, believes that we will see increasing levels of autonomous driving arriving in the coming months.

Speaking to Top Gear during a test drive of the upcoming R2, which the company hopes will be its first electric SUV with true mass appeal, Scaringe revealed that he thinks we will move from level two to three, which includes hands-off and eyes-off autonomous driving, within “the next 18 months”.

A new Leiden Declaration, endorsed by the International Mathematical Union and published on June 2, 2026, warns that AI could undermine mathematics by flooding the field with plausible but flawed proofs, weakening attribution, shifting incentives, and giving tech companies too much influence over research priorities. “Mathematicians should find it quite striking that tech companies are suddenly interested in their work,” said Kevin Buzzard, a mathematician at Imperial College London, in a statement. “The Leiden Declaration is a well-thought-through response to what is currently happening, as AI continues to disrupt this space.” Ars Technica reports: The Leiden Declaration, which has already drawn hundreds of signatories, warns that recent AI developments are threatening “characteristic values” of mathematical research, “often in ways that disproportionately affect students and early-career mathematicians, and hence the long term future of the discipline.”

First, it points out how AI models can “produce plausible but unreliable (or even incorrect) arguments which are difficult to distinguish from correct mathematical proofs.” Such developments put reviewers under increasing pressure and are “jeopardizing our ability to implement traditional standards for the correctness, transparency, and independent verifiability of proof,” the declaration warns. “Inaccurate AI-generated drafts are cheap to produce, and there is a risk of cluttering the literature with claimed results that are simply wrong,” said Leslie Ann Goldberg, head of computer science at the University of Oxford, in a statement. “Once that happens, the errors are likely to propagate as new results are built on faulty foundations.”

Second, the declaration highlights how “models trained on published works frequently return outputs that do not properly cite the human works they synthesize,” while also pointing out that many current AI models were trained on data obtained through “exploiting licenses and access arrangements” or “simply violating copyright protections.”

Third, the declaration describes how the use of AI “may become incentivized for its own sake, disrupting our mechanisms for hiring, funding and recognition” while leaving out researchers who lack access or are “unwilling to use technologies controlled by organizations whose values they do not share.”

Fourth, the declaration warns against mathematics research “communicated through informal channels such as press releases or blog posts, often without any research paper or other disclosure of information necessary for scientific evaluation.” Such communication strategies can lead to “oversimplification” in media reporting that overemphasizes AI tools’ significance at the expense of prior human contributions, and “misleadingly uses specific mathematical tasks as metrics for the general reasoning capacities of commercial products.”

Fifth, the declaration describes “increasing involvement of technology companies in mathematical research” as threatening the “autonomy of mathematics,” especially as university budgets are under pressure and researchers may feel greater professional incentive to collaborate with technology companies on “asymmetric terms.” This also raises the risk that mathematics research questions amenable to AI-driven techniques may be prioritized. What can mathematicians do about this? The Leiden Declaration urges them to treat AI as a tool, not a substitute for human responsibility. Individual mathematicians should disclose AI use, remain accountable for the correctness of their work, continue crediting human authors, and use AI tools only when they align with the declaration’s values.

It also warns that mathematics can be applied to “warfare, oppression, mass surveillance, and the undermining of democracy,” so mathematicians should weigh the ethics of tech-industry partnerships carefully. Professional organizations are encouraged to develop AI-use guidelines for publication and review, protect researchers from having their work used as training data without consent, support peer-reviewed publishing, and “actively prepare to become involved if major mathematical results are claimed using unconventional means.”

For policymakers, the recommendations are blunt: “protect the rights of authors,” “regulate the artificial intelligence industry,” and “invest in public computational infrastructure.” The declaration also urges people to “don’t believe the hype,” warning that tech companies have “a strong commercial incentive… to overstate the capabilities of their products.”

• Microsoft CEO looks to ease data center environmental fears
• A “new approach” will help the company’s facilities address concerns, Nadella says
• Microsoft’s Azure cloud business now covers more than 500 data centers in 80 regions

Microsoft CEO Satya Nadella has looked to reassure those concerned about the effect data centers are having on the environment.

Speaking during his keynote address at Microsoft Build 2026, Nadella outlined how the company is working on “a new approach” to its data centers, with plans to improve cooling systems and reduce water use

• Email systems were never designed for autonomous machine workflows
• Hostinger introduces webhook-first email for real-time automation processing
• AI agents now trigger actions immediately when emails arrive

AI agents can process data and execute actions within milliseconds, yet many automated systems still depend on tools originally built for human users.

That mismatch has become increasingly noticeable as businesses attempt to connect AI-driven workflows with traditional email systems, never designed for machine-to-machine interactions.