Open

Category Archives: Tech

Tamara News  /  Blog  /  Tech
22 Sep 2026
European Union flag in Brussels amid EU AI Act compliance audits

Brussels Just Sent Its First Round of AI Audit Letters — Here’s Who Got One

Categories Tech, World

The European Commission’s AI Office has opened its first wave of formal compliance checks under the bloc’s AI Act. Sending information requests to more than 30 AI model providers. These EU AI Act compliance audits mark the first concrete use of the law’s investigative powers since high-risk obligations became applicable in August.

Compliance documents reviewed during EU AI Act compliance audits

How the EU AI Act compliance audits are structured

The European AI Office in Brussels is working alongside 24 national market surveillance authorities on the first scheduled inspection wave. France’s CNIL, Germany’s BfDI and Spain’s AESIA are focusing their initial requests on three specific sectors. Automated resume screening tools used in human resources, algorithmic credit assessment systems in retail banking. And aI triage tools deployed in private healthcare clinics. Those three sectors were chosen because they combine high-risk classification under the Act with large numbers of affected consumers.

Regulators are examining whether providers maintained the technical documentation required under Article 11 and Annex IV of the Act. That includes version-controlled model weights, pre-processing scripts and detailed descriptions of human-override mechanisms built into each system.

Which AI systems fall under scrutiny

The scope of the audits covers systems deployed after 2 August 2026. The date high-risk obligations under the Act became fully applicable. Any model released or meaningfully updated since that date in one of the three targeted sectors is now subject to inspection. Providers outside those three sectors are not exempt long-term. Regulators have signaled this is the first of several planned waves rather than a one-time exercise.

Legal advisers working with AI vendors say the sector selection was deliberate. Resume screening, credit scoring and medical triage all involve automated decisions that directly affect individual consumers’ access to jobs. Loans and healthcare, making them the clearest test cases for whether the Act’s human-oversight requirements function in practice rather than only on paper. A weak response from providers in these three sectors would likely accelerate scrutiny of AI systems in other consumer-facing industries sooner than originally planned.

How this differs from August’s transparency rules

This audit wave is distinct from the transparency requirements that took effect in late August. This focused on disclosure obligations for general-purpose AI models regardless of sector. The current compliance checks go further, testing whether companies can actually produce the technical documentation the law requires. Not just whether they have published the right disclosures. Companies that met the August transparency deadline are not automatically in the clear for this new documentation-focused review.

That distinction matters for compliance planning. A company can publish a fully compliant transparency notice while still lacking the underlying version-controlled documentation regulators are now requesting. That is because the two requirements test different things. What a company tells the public versus what a company can actually prove internally. Companies that treated the August deadline as the finish line for EU AI Act compliance are now discovering it was closer to a starting point.

Compliance consultants working with mid-size AI vendors report a scramble in the days since the requests went out. As legal and engineering teams try to reconstruct documentation for models that were built and shipped quickly during the competitive rush of the past two years. For companies that treated documentation as an afterthought while racing to deploy. The audit letters have turned into an unplanned. Resource-intensive project with a regulator-set deadline rather than an internal one.

Smaller AI startups face a particular bind. Unlike the largest labs. Many lack dedicated compliance staff and had assumed enforcement would focus first on the biggest. Most visible model providers. The decision to open initial audits within three sector-specific use cases rather than by company size caught some smaller vendors in HR-tech and health-tech off guard. That is because a startup with a single flagship product in one of these three sectors now faces the same documentation demands as a much larger competitor.

What happens next for audited companies

Companies that receive information requests typically have a defined window to respond with documentation before regulators decide whether to escalate to a formal investigation. The EU AI Act allows for significant fines for non-compliance, scaled to global revenue for the largest providers. Given that this is the first wave of a stated multi-wave process. How these initial 30-plus companies respond is likely to shape enforcement patterns for every subsequent round. That includes in sectors beyond the three targeted so far.

Frequently asked questions

Who is conducting the EU AI Act compliance audits?
The European Commission’s AI Office in Brussels, working with 24 national market surveillance authorities including France’s CNIL, Germany’s BfDI and Spain’s AESIA.

Which sectors are being audited first?
Automated resume screening in HR, algorithmic credit assessment in retail banking, and AI triage tools in private healthcare clinics.

What documentation are regulators requesting?
Technical files required under Article 11 and Annex IV, including version-controlled model weights, pre-processing scripts and descriptions of human-override mechanisms.

Which AI systems are covered?
Systems deployed after 2 August 2026, when high-risk obligations under the EU AI Act became fully applicable.

Is this different from the August transparency rules?
Yes. The August rules covered disclosure obligations for general-purpose models, while this audit wave tests whether companies can produce required technical documentation.

Related coverage

Sources

22 Sep 2026
Courtroom gavel symbolizing the AI antitrust lawsuit

Four People Are Suing Anthropic, OpenAI, Google and SpaceXAI — For Agreeing to Slow Down

Categories Business, Tech

Four paying subscribers to leading AI chatbots have filed a federal antitrust lawsuit against Anthropic. OpenAI, Google and SpaceXAI. This AI antitrust lawsuit makes an unusual claim: the companies agreed to slow down their own technology. And that agreement broke the law. The case, filed last week, argues the pact reduced the value subscribers get from their paid plans.

AI server data center at the center of the AI antitrust lawsuit

What triggered the AI antitrust lawsuit

The dispute traces back to 12 September. When Anthropic chief executive Dario Amodei published an essay urging leading AI firms to cooperate on slowing the pace of capability advances. Amodei framed it as a safety measure. His plan, sometimes called “pacing the frontier,” called for outside evaluators embedded inside AI companies and a shared. Industry-wide agreement on safety benchmarks and how fast capabilities should be allowed to advance.

On the same day, OpenAI chief executive Sam Altman, SpaceXAI chief executive Elon Musk. And google DeepMind co-founder Demis Hassabis each responded in public support. That coordinated agreement is now the basis of the lawsuit. The plaintiffs argue the four firms improperly restrained the pace of their own technology’s progress. This they say violates federal antitrust law.

The plaintiffs’ central argument

According to court filings reported by CNN and CBS News. The four subscribers claim the slowdown pact undermines the value they receive from their paid subscriptions. Legal filings allege that industry-wide coordination on capability pacing began months before Amodei’s public essay. That means the September statements may have been a public confirmation of private discussions rather than the start of the agreement itself.

Altman has since said OpenAI would welcome a consistent federal safety framework. But he added that the company did not believe it needed to wait for an antitrust waiver or new legislation before adopting new safety practices on its own. That response suggests the four companies may not present a unified defense as the case proceeds.

Why “safety cooperation” and antitrust law collide

Amodei’s original proposal included a request. He suggested the US government grant AI companies a restricted waiver specifically permitting safety-related discussions that would otherwise raise competition concerns. That request itself is a tell. Ordinarily, competitors coordinating on output, pricing, or the pace of product development invites antitrust scrutiny. This holds regardless of the stated motive. Whether “capability pacing” counts as a legitimate safety measure or an anticompetitive restraint on trade is now a question for the courts rather than the companies themselves.

Legal scholars who study antitrust law say the case sits in genuinely unsettled territory. Companies routinely coordinate on safety standards in other industries, from aviation to pharmaceuticals. Without triggering antitrust liability, provided the coordination does not extend to output or pricing. The plaintiffs’ argument is that slowing capability development functions economically like restraining output. That is because capability improvements are effectively the product AI subscribers are paying for. Whether a court accepts that framing will likely determine how the case proceeds from here.

What happens next for the four AI companies

None of the four companies has yet filed a formal response in court. The case adds a new front to a year that already saw Anthropic sued separately by Sony and Warner over AI music training. And increased federal attention on chip export rules and frontier model safety commitments. Antitrust cases against technology companies often take years to resolve. But early procedural rulings on whether the “pacing the frontier” agreement counts as coordination among competitors could shape how AI firms discuss safety cooperation going forward.

The broader industry is watching closely, too. If courts side with the plaintiffs. Any future joint safety statement from competing AI labs could become a liability risk rather than a public-relations win. That would push safety coordination toward government-brokered frameworks, the kind of arrangement Amodei originally asked for. It is not informal agreements announced through public essays and social media posts.

Consumer advocacy groups have taken a different view of the case, arguing that regardless of the antitrust technicalities. The lawsuit surfaces a legitimate question about who gets to decide how fast AI capabilities advance. Subscribers, the argument goes, are paying for continuous improvement. And a private agreement among four companies to slow that improvement was never put to a vote or a public comment process. Whether that framing carries any legal weight is separate from whether it resonates politically. And early commentary suggests it already has.

Frequently asked questions

Who filed the AI antitrust lawsuit?
Four paying subscribers to AI chatbot services filed the suit against Anthropic, OpenAI, Google and SpaceXAI.

What is “pacing the frontier”?
It is Dario Amodei’s proposal for AI companies to coordinate on slowing capability advances, using outside evaluators and shared safety benchmarks.

Why do the plaintiffs say this violates antitrust law?
They argue the companies improperly agreed to restrain the pace of their technology, which reduced the value subscribers receive from paid plans.

Did all four companies support the slowdown plan?
Altman, Musk and Hassabis each publicly responded in support of Amodei’s proposal on 12 September, though Altman later said OpenAI would not wait for legal cover to adopt new safety practices.

Has any company responded formally to the lawsuit?
No formal court response had been filed by any of the four companies as of this report.

Related coverage

Sources

21 Sep 2026
Data center hardware representing the StepFun Step 5 Preview AI model launch

A Chinese Startup Just Undercut OpenAI on Price by a Wide Margin

Categories Tech

Chinese AI lab StepFun launched its flagship model on September 20. The StepFun Step 5 Preview is already turning heads for one reason as much as any benchmark: price.

The 600-billion-parameter sparse mixture-of-experts model went live with API pricing of $1.00 per million input tokens. Output runs $2.70 per million tokens. That undercuts several Western frontier labs by a wide margin. It still scores competitively on independent benchmarks.

Server room hardware powering the StepFun Step 5 Preview AI model

What the StepFun Step 5 Preview Actually Is

Step 5 Preview carries roughly 600 billion total parameters. About 27 billion activate per token. That sparse architecture keeps inference costs down while preserving capacity. It supports a one-million-token context window. It takes both text and image inputs. The design targets long, complex tasks rather than short chat exchanges.

How It Stacks Up Against the Field

Step 5 scored 44 on the widely tracked Intelligence Index. That matches Kimi K3 Max, another Chinese frontier model. In one benchmark test, it completed a 24-hour autonomous optimization task. It ran at 508 TFlops on H100 GPU cores. StepFun points to that result as evidence the model can sustain long, agentic workloads without human help.

Matching a well-regarded rival like Kimi K3 Max does not by itself prove Step 5 leads the field. But it places StepFun firmly among the top tier of Chinese labs. Those labs now compete directly with the largest Western AI companies on capability, not just cost.

The Pricing Move That Has Rivals Watching

StepFun built the release around long-horizon agent workloads. Think AI coding, software engineering, financial analysis and professional knowledge work. Pairing that focus with aggressive pricing puts direct pressure on Western labs. Enterprise customers running high-volume agentic tasks watch token costs closely. Those costs add up fast at scale.

Open Weights Land October 15

Unlike many frontier releases that stay fully closed, StepFun has committed to opening the model’s weights on October 15, 2026. That gives outside researchers and companies a firm date to plan around. It continues a pattern of Chinese labs using open weights as a lever against closed Western rivals.

Why Chinese Labs Keep Undercutting on Price

StepFun’s pricing move follows a broader pattern among Chinese AI labs this year. Several have used aggressive token pricing to win developer mindshare. That holds even when their raw benchmark scores trail the very top Western models. The strategy leans on a simple bet. Once developers build workflows around a cheaper model, switching costs make it harder for pricier rivals to win them back.

That approach has drawn scrutiny from analysts. They question whether the pricing holds up at scale. For now, though, it has pulled models like Step 5 into serious enterprise evaluations. Those evaluations might otherwise have gone straight to established Western providers.

Where StepFun Fits Among Chinese AI Labs

StepFun is one of several Chinese AI labs that have released a series of models under a numbered naming scheme, steadily scaling parameter counts with each generation. Step 5 Preview is the company’s most ambitious release to date, both in size and in the workloads it targets.

Unlike rivals that lead with chat or search products, StepFun has positioned this release around agentic and enterprise workloads from the start. That focus shapes how the company talks about the model. Its own materials emphasize coding and financial-analysis benchmarks over general conversational ability.

What This Means for Enterprise AI Buyers

Enterprise teams evaluating AI vendors now have another low-cost option to weigh alongside the established players. Procurement teams say token pricing has become a bigger factor in vendor selection as usage scales into the billions of tokens monthly. A cheaper model that performs adequately can save real money at that volume, even if it lags top benchmarks slightly.

Data residency and compliance concerns may still steer some Western enterprise buyers away from Chinese-hosted models, regardless of price. That question is likely to shape how far StepFun’s pricing advantage actually translates into enterprise contracts outside China.

What to Watch Before the Weights Drop

The real test comes once independent developers get hands-on access to the open weights next month. Benchmark scores from a lab’s own team invite scrutiny. The coming weeks will show whether Step 5’s real-world coding and agent performance holds up outside StepFun’s own testing environment.

Until then, the API launch already lets outside developers test the model directly. That should produce a clearer, independent picture of its performance well before the October 15 open-weights date.

Straight Answers on StepFun’s New Model

What is the StepFun Step 5 Preview?

A 600-billion-parameter sparse mixture-of-experts AI model from Chinese lab StepFun, launched September 20, 2026, with about 27 billion parameters active per token.

How much does it cost to use?

API pricing is $1.00 per million input tokens and $2.70 per million output tokens.

What is it designed for?

Long-horizon agent workloads such as AI coding, software engineering, financial analysis and professional knowledge work.

How big is its context window?

One million tokens, with support for both text and image inputs.

When do open weights arrive?

StepFun has scheduled the model’s weights to open on October 15, 2026.

Sources

More AI coverage on Tamara News: OpenAI just admitted its models are lying and hiding mistakes and Perplexity’s new AI agent wants to run your PC for you.

21 Sep 2026
Semiconductor fab floor tied to CXMT G5 DRAM production in China

China’s Chip Champion Just Closed the Gap on Samsung and SK Hynix

Categories Tech

CXMT G5 DRAM production is now underway in China. It marks one of the clearest signs yet that the country’s chipmakers are closing the gap with Samsung and SK Hynix. That progress comes despite years of export restrictions on advanced lithography gear.

Chinese memory chipmaker CXMT announced the milestone at the 2026 World Manufacturing Convention in Hefei on September 20. It unveiled its fifth-generation DRAM platform, known as the G5 Platform.

Close-up of a memory chip representing CXMT G5 DRAM production

What CXMT G5 DRAM Production Actually Achieves

The G5 Platform uses quadruple patterning. That reaches an 11.95-nanometer active-area half-pitch for memory arrays. CXMT said its capacitor aspect ratio reaches 45:1. The core cell array height has shrunk to 6,762 nanometers. In plain terms, the company packs more memory into a smaller footprint. It says the platform yields at least 50% more dies per wafer than its predecessor.

How CXMT Got Here Without ASML’s Banned Machines

The achievement stands out for one reason. CXMT built it without access to ASML’s most advanced extreme ultraviolet lithography tools. Those remain subject to export controls. The company instead relied on quadruple patterning with existing equipment. That workaround took years to refine. It appears to be paying off at scale now.

Multiple patterning is not new in chipmaking. But matching the density gains EUV lithography delivers in a single pass usually costs more. It means more process steps, more time and more cost per wafer. CXMT chose to absorb that complexity rather than wait for policy changes. That choice reflects the broader bet Chinese chipmakers have made since export controls tightened.

What This Means for Samsung, SK Hynix and Global Memory Prices

CXMT now claims the title of China’s largest DRAM supplier. It also claims fourth place worldwide by market share. The company has scaled fast. It grew from roughly 40,000 wafers a month in 2020 to about 300,000 today. That output spans its 12-inch fabs in Hefei and Beijing. It targets 350,000 to 375,000 wafers monthly by year-end 2026. That capacity growth adds real supply to a global memory market squeezed by AI-driven demand.

The Products Already Shipping on the New Platform

Two 24-gigabit LPDDR5X products are already in mass production. They come in 496-ball and 245-ball packages for smartphones and portable electronics. Both are integrated into mainstream Chinese flagship handsets already, the company says.

The Bigger Picture for China’s Chip Ambitions

CXMT’s progress fits a wider Chinese push to build a self-sufficient semiconductor supply chain. That priority hardened after successive rounds of US export controls on advanced chipmaking equipment. Beijing has funneled heavy state-backed investment into domestic memory and logic chip production for years. CXMT’s rise from a minor player to a top-four global DRAM supplier is one of the clearest results so far.

Whether the company can keep closing the gap at the cutting edge remains an open question. Both Korean rivals keep investing heavily in their own next-generation nodes. Neither shows signs of slowing down.

Which Chinese Phone Makers Are Buying In

CXMT has not publicly named every customer using the new LPDDR5X chips. Industry watchers tracking teardown reports say several major Chinese smartphone brands have already qualified the parts for mainstream, non-flagship models. That matters commercially. Winning approval in China’s huge domestic phone market gives CXMT steady volume, even before it competes for premium international designs.

Analysts say the bigger prize is qualification with global device makers outside China. That would require CXMT to match not just density but also long-term reliability data that Samsung and SK Hynix have built up over decades. CXMT has not said when, or whether, it plans to pursue that qualification process.

Where CXMT Goes From Here

Capacity is still climbing toward its year-end target. CXMT’s next test is sustaining yields at scale while narrowing the technology gap further. Industry watchers expect the company’s progress to keep pressuring global memory pricing. That dynamic is already squeezing device makers worldwide this year.

The next milestone to watch is server and data-center memory. Can CXMT push the G5 platform into that higher-capacity segment? It remains dominated even more heavily by Samsung, SK Hynix and Micron than the mobile market CXMT has targeted so far. Analysts see that as the real test of whether CXMT can compete at the very top of the market, not just at the budget end.

Quick Answers on China’s New Memory Chip

What is CXMT G5 DRAM production?

It is the mass production of CXMT’s fifth-generation DRAM platform, announced at the 2026 World Manufacturing Convention in Hefei on September 20.

How advanced is the new chip?

The G5 Platform achieves an 11.95-nanometer active-area half-pitch using quadruple patterning, without relying on ASML’s most advanced banned lithography machines.

What products use it already?

Two 24-gigabit LPDDR5X chips, packaged for smartphones and portable electronics, are already in mass production and integrated into mainstream Chinese flagship phones.

How much is CXMT producing?

The company has grown from roughly 40,000 wafers per month in 2020 to about 300,000 wafers per month across its Hefei and Beijing fabs, targeting 350,000 to 375,000 by year-end 2026.

Where does CXMT rank globally?

CXMT now claims to be China’s largest DRAM supplier and the world’s fourth-largest by market share.

Sources

Related coverage on Tamara News: your next phone costs more because AI bought the memory first and Nvidia’s $12.9B Hugging Face deal.

20 Sep 2026
Trump AI czar announcement artificial intelligence and technology concept

Trump Wants an ‘AI Force.’ He Hasn’t Said Who’s Flying It.

Categories Tech, World

President Trump used a Saturday Truth Social post to announce two things at once: a new “AI Force,” modeled on the Space Force he created in his first term, and plans to name an AI czar to oversee it. The Trump AI czar announcement came on September 19, with Trump adding that “only High I.Q. individuals need apply” for the role.

No name, timeline or budget accompanied the post. The announcement came days after AI researchers renewed public warnings that advanced systems could become difficult to control, a debate Trump has previously dismissed as overblown.

The post was short. The implications were not. A single paragraph on social media now sits at the center of a debate over how the US government will treat the fastest-moving technology of the decade.

The short version: Trump posted the plan on September 19. It creates an “AI Force” modeled on Space Force. An AI czar will lead it. Trump has not named anyone yet. Trump still calls AI fears a hoax. He says enforcement will target “BAD” actors.

What Trump Actually Announced

The post described an “AI Force” without detailing its legal authority, funding source or relationship to existing agencies that already touch AI policy, including the Commerce Department’s semiconductor export controls and the President’s Council of Advisors on Science and Technology. Trump said an AI czar would come “in the near future” to take point on the effort.

Trump cited no bill. No agency memo followed. Reporters asked for detail. The White House had little more to add. That silence is itself part of the story.

The Contradiction at the Center of the Announcement

Trump has repeatedly called fears about AI risk a “hoax,” a position he has held publicly for months. Yet this announcement followed renewed warnings from AI researchers about the technology moving faster than oversight can keep up with. Trump appeared to partially acknowledge this tension, saying his administration would go after “BAD” actors using existing criminal and civil justice tools rather than new AI-specific regulation.

Who Held This Role Before

Venture capitalist David Sacks previously served as Trump’s AI and cryptocurrency czar before stepping down in March 2026 after reaching the time limit allowed for a special government employee. Sacks now chairs the President’s Council of Advisors on Science and Technology, meaning any new AI czar would be working alongside, not replacing, an official already positioned at the center of the administration’s tech policy.

That overlap raises a practical question the announcement did not address: whether an AI czar reporting through a new AI Force structure would have authority over, alongside, or beneath the existing science and technology advisory council Sacks now leads.

How the Trump AI Czar Announcement Landed With Researchers

The timing struck many observers as notable. Researchers across multiple AI labs had spent the preceding week publicly renewing warnings about the pace of frontier AI development outstripping safety testing, a debate that has grown louder throughout 2026. Trump’s announcement arrived days later, framed not as a response to those warnings but as a parallel initiative focused on enforcement against bad actors rather than restrictions on model development itself.

That framing puts the administration’s position somewhere between the AI industry’s largest labs, several of which have called for more structured oversight, and skeptics in Congress who have resisted new AI-specific regulatory bodies. Where the proposed AI Force ultimately lands on that spectrum will depend heavily on who is named to lead it.

What We Still Don’t Know

The announcement left the most consequential questions open: what authority the AI Force would actually have, how it would be funded, whether it requires congressional action, and who Trump will name to lead it. Until those details arrive, the announcement functions more as a signal of intent than a concrete policy shift.

Congressional committees with jurisdiction over technology policy have not yet indicated whether they were consulted before the announcement or whether hearings on the proposal are planned.

The numbers at a glance: Announcement date: September 19. Platform used: Truth Social. Model cited: Space Force. Named leader: none yet. Previous AI czar: David Sacks. Sacks stepped down: March 2026.

What People Are Asking About the AI Force

What is Trump’s proposed “AI Force”?
A new body Trump says will be modeled on the Space Force to oversee artificial intelligence, announced without further structural detail.

When was the Trump AI czar announcement made?
September 19, 2026, via a post on Truth Social.

Who is expected to be named AI czar?
Trump has not named a candidate. He said an announcement would come “in the near future.”

Who held a similar role before?
David Sacks served as AI and cryptocurrency czar until March 2026, when he reached his time limit as a special government employee.

Does this mean new AI regulation is coming?
Not necessarily. Trump indicated the administration would pursue “BAD” AI actors through existing criminal and civil law rather than new AI-specific rules.

Recommended Reading

Sources

  • NBC News — Trump Says He’s Creating an AI Force and Appointing a Czar Amid Concerns Over the Rapidly Developing Tech. nbcnews.com
  • CNN — Trump Vows to Create ‘AI Force’ and Appoint Czar Amid Calls for Oversight. cnn.com
  • Al Jazeera — Trump Says He Will Create ‘AI Force’ With New ‘AI Czar’. aljazeera.com

1 … 9 10 11 12 13 … 26