• Mon - Fri 9.00 - 19.00

Yearly Archives: 2026

Renewing an H-1B Just Got $4,000 Costlier, For Some Employers

A new US rule took effect on 9 September 2026 that adds thousands of dollars to certain work visa renewals. The H-1B extension fee now reaches $4,000 for covered employers, and the equivalent L-1 charge reaches $4,500. The change applies to extension petitions that previously escaped it.

The Department of Homeland Security published the final rule on 10 August 2026. It amends the regulations governing the 9-11 Response and Biometric Entry-Exit Fee.

Form I-129 petitions postmarked or filed electronically on or after 9 September must include the fee where it applies.

Who the H-1B extension fee actually hits

The charge is limited to “covered employers”. The definition has two parts and both must be true.

The employer must have 50 or more US employees. More than 50% of that workforce must hold H-1B or L-1 status.

Hand typing on a keyboard representing employers affected by the H-1B extension fee

That test excludes most companies outright. A large firm with a handful of visa holders does not meet it. A mid-sized firm built mainly on sponsored staff does.

The rule therefore concentrates on staffing and outsourcing businesses rather than on employers generally. Most H-1B and L-1 workers will not see their employer charged.

What changed on 9 September

The fee itself is not new. It already applied to petitions seeking an initial grant of status, and it still does, unchanged.

What changed is the scope. Covered employers must now pay it on all extension of status petitions.

Previously the charge was tied to whether the related fraud prevention and detection fee applied. The new rule removes that link. Extensions that keep an employee with the same employer are now captured.

That last point matters most in practice. Routine renewals for existing staff were the common case that fell outside the old wording.

What this means for workers on these visas

The fee is an employer cost. Workers do not pay it directly.

Indirect effects are the real question. Employers facing a higher renewal cost may weigh sponsorship differently at the point of extension, particularly where margins are thin.

Consider a software engineer on an H-1B at a mid-sized consultancy where most staff hold sponsored status. That employer now faces $4,000 at each extension on top of existing filing costs. The engineer’s status is unaffected, but the renewal decision carries a new number.

Nothing in the rule changes eligibility, processing times or the cap. It changes price, and only for a defined set of employers.

This sits alongside other cost changes we have covered, including the proposed six-figure H-1B fee and the weighted selection and prevailing wage proposal.

How the 50-50 test works in practice

The threshold has a specific history. Congress created the surcharge for employers whose US workforce depends heavily on H-1B and L-1 staff, and the 50-50 label comes from its two limbs.

Counting the workforce is the first step. Employers count US employees, including full-time and, depending on their policy, other categories their counsel treats as employees.

Counting visa holders is the second. More than half of that total must hold H-1B or L-1 status for the surcharge to apply.

Employers near the line need care. A company at 48% one quarter and 52% the next changes category, and the test applies at the time of filing.

Getting this wrong cuts both ways. Underpaying risks rejection of the petition. Overpaying hands money to an agency that will not volunteer it back.

Dates and filings to keep in view

The 9 September trigger is based on the postmark or electronic submission date, not on the decision date. Petitions sent before then follow the old rules.

Employers close to the 50-employee or 50% thresholds should confirm their position before filing. The test is applied to the employer, not the individual petition.

Separately, a revised public charge guidance applies to Form I-485 filings postmarked on or after 18 September 2026. Applications filed before that date fall under the 2022 policy. That is a different rule affecting a different filing.

September is also the final month of the US fiscal year, which is when employment-based categories can hit annual limits. Our summary of the October 2026 Visa Bulletin covers where those categories stand.

Keep the two filings separate in your planning. An I-129 extension and an I-485 adjustment follow different rules, different fees and different cut-off dates.

Employers running both for the same worker should date each one deliberately. A single week can move a filing from one rule set to another.

Frequently raised points on the new charge

  • How much is the fee? $4,000 for H-1B and $4,500 for L-1, payable by covered employers.
  • When did it start? The final rule took effect 9 September 2026 and applies to Form I-129 petitions filed on or after that date.
  • Who counts as a covered employer? An employer with 50 or more US employees where more than 50% hold H-1B or L-1 status.
  • Does it apply to new petitions? The fee already applied to initial grants of status and continues to apply there unchanged.
  • What exactly changed? Covered employers must now pay it on all extension petitions, including extensions with the same employer.
  • Do employees pay it? No. It is charged to the petitioning employer.

Related immigration coverage

Read our reports on the $100,265 H-1B fee proposal, the weighted H-1B selection plan and the October 2026 Visa Bulletin.

Sources

  • Federal Register — 9-11 Response and Biometric Entry-Exit Fee for H-1B and L-1 Visas. federalregister.gov
  • American Immigration Lawyers Association — DHS Final Rule Expanding 9-11 Response and Biometric Entry-Exit Fee. aila.org
  • USCIS — H-1B Specialty Occupations. uscis.gov

OpenAI Just Gave Every Developer the Tools Behind Its Coding Agent

OpenAI opened public beta access to its new Agents API on September 10, 2026, expanding what outside developers can build with its technology. It puts the infrastructure behind its Codex coding agent directly into developers’ hands through a single API call. The OpenAI Agents API lets outside developers build custom AI agents that can run code, browse the web and coordinate their own subagents. They no longer need to build that scaffolding themselves from the ground up.

Software development workspace representing the new OpenAI Agents API

What the OpenAI Agents API Actually Does

According to OpenAI’s own announcement, the API brings the same harness and infrastructure that powers Codex to any developer through a flexible interface. Inside a running session, an agent built on the API can execute code, edit files, search the web, apply predefined skills, and produce artifacts. It can also split tasks across subagents up to a set concurrency limit. Those capabilities previously required developers to assemble their own agent infrastructure from scratch.

How the Architecture Is Built

The API is organized around four core objects, as detailed by MarkTechPost’s technical breakdown. An Agent object defines the model, instructions, tools and any connected MCP servers. An optional Environment sandbox handles isolated execution. A durable Session persists state, and a stream of events reports what the session is doing as it works. Developers can run that compute inside an OpenAI-managed sandbox, on their own infrastructure, or through a partner sandbox. Integrations are already live from Cloudflare, DigitalOcean and Oracle.

Why This Matters Beyond Coding Tools

Codex itself remains a specific product aimed at software development. Exposing its underlying harness as a general-purpose API changes that. Developers outside the coding-tools space can now build agents for customer support, research or data analysis. Any workflow that benefits from an AI system capable of taking multi-step action, not just answering questions, is now a candidate. That positions OpenAI more directly against agent-infrastructure offerings from rivals racing to give developers similar building blocks.

Developer writing code that could integrate with the new OpenAI Agents API

What Comes Next in the Beta

The Agents API launched with pay-per-use pricing based on tokens and tool calls. There is no separate subscription layered on top. As the public beta continues, the real test will be adoption. Will third-party developers use the harness at scale for production workloads? And how quickly will OpenAI expand its list of managed-sandbox and infrastructure partners beyond the three announced at launch?

How This Fits the Wider Agent Race

OpenAI is not the only company racing to give developers agent-building infrastructure. Anthropic offers its own agent development kit for building custom Claude-based agents, and Google has been expanding similar tooling around its Gemini models. What sets OpenAI’s move apart is that it exposes the exact harness already proven in a shipping product, Codex. That is different from a separate framework built specifically for external developers. Developers weighing whether to adopt the new API will likely focus on pricing predictability and how easily existing Codex-based workflows can migrate over without a rewrite. OpenAI has not said whether Codex itself will eventually be rebuilt on top of the public Agents API or continue as a separate, parallel product. Infrastructure partners such as Cloudflare and DigitalOcean stand to benefit either way. Developer demand for sandboxed agent compute tends to translate directly into cloud spending, regardless of which underlying model or harness an agent uses.

Frequently Asked Questions

What is the OpenAI Agents API?

It is a public beta, launched September 10, 2026, that gives developers the same harness and infrastructure that powers OpenAI’s Codex coding agent. It is accessible through a single API rather than a standalone product.

What can developers build with it?

Inside a session, the Agents API can run code, edit files, search the web, apply skills and produce artifacts. It can also split work across subagents with a set concurrency limit, according to OpenAI’s own announcement.

How is the API structured?

It is built around four objects. An Agent defines the model, instructions, tools and MCP servers. An optional Environment sandbox handles execution, a durable Session persists state, and a stream of events reports what the session produces as it runs.

Where does the agent’s compute actually run?

Developers can choose to run agent compute in an OpenAI-managed sandbox, their own infrastructure, or a partner sandbox. Integrations are already available from Cloudflare, DigitalOcean and Oracle.

How much does it cost to use?

The Agents API operates on a pay-per-use model based on tokens and tool usage, with no additional flat fee on top of standard API costs.

Is this different from OpenAI’s Codex product?

Codex remains a specific coding-agent product. The Agents API exposes the same underlying harness so developers can build their own custom agents for tasks well beyond coding.

The launch adds to a busy stretch for OpenAI, which has spent September rolling out new capacity, safety governance changes and now a broader developer platform in quick succession. Developers who build on the new API will be watching closely for how OpenAI balances rapid feature releases against reliability. Enterprise customers expect stability from production infrastructure as more companies move agent workloads from experiments into everyday business use. For more on the AI infrastructure race this month, see our coverage of OpenAI’s GPT-6 Astra release and the memory chip shortage squeezing every major AI hardware maker.

Sources

The DOJ Wants to Know Why Nvidia’s $20B Groq Deal Wasn’t a Merger

The US Justice Department is examining whether Nvidia structured a $20 billion licensing agreement with AI chip startup Groq to avoid antitrust review. The Nvidia Groq antitrust probe was reported on 10 September 2026. Investigators have sent Nvidia a formal demand for information.

Groq presented the December arrangement as a non-exclusive licence. It gave Nvidia access to chips optimised for AI inference.

Two Groq executives moved to Nvidia as part of the same deal. Founder and chief executive Jonathan Ross and chief operating officer Sunny Madra both joined. Groq remained a separate company.

What the Nvidia Groq antitrust probe is testing

Merger review in the United States is triggered by acquisitions above certain thresholds. A licence is not an acquisition. Neither is hiring executives.

The question is whether those pieces together achieve what an acquisition would, without the filing that an acquisition requires. Regulators have looked at similar structures across the AI sector over the past two years.

Server room representing the AI inference capacity at stake in the Nvidia Groq antitrust probe

The department opened the probe soon after the arrangement was announced, according to the reporting. That timing suggests the structure itself drew attention rather than any later conduct.

Why inference chips are the contested ground

Training a model and running it are different workloads. Training happens once and demands enormous parallel compute. Inference happens every time someone uses the model.

Inference volume grows with adoption. That makes inference silicon the larger long-run market, and the one where challengers have had the clearest opening against Nvidia.

Groq built its business on that opening. A deal that gives Nvidia access to its inference-optimised designs while moving its leadership across narrows the field, at least on paper.

None of that establishes a violation. A formal demand for information is an investigative step, not a finding.

The acqui-hire pattern regulators keep meeting

The structure at issue has become common across artificial intelligence. A large company licenses a startup’s technology, hires its senior leadership, and leaves the startup standing as a separate entity.

Each element looks ordinary on its own. Licences happen constantly. Executives change jobs constantly. Neither triggers a merger filing.

Combined, they can transfer the two things that matter most: the technology and the people who built it. What remains at the startup is a name, a cap table and whatever staff stayed behind.

Regulators in the United States, the United Kingdom and the European Union have all examined arrangements of this shape since 2024. Outcomes have varied, and several closed without action.

That history explains the department’s interest here without predicting where it lands. Structure alone does not establish harm, and the law asks about effects on competition rather than about form.

How this fits Nvidia’s wider regulatory picture

Nvidia is under pressure on several fronts at once. Export controls shape what it can sell and where, a subject we covered in our report on the chip export loophole.

Supply is the second front. Our note on the memory chip shortage sets out the constraint running through the whole accelerator market.

Litigation across AI companies is the third. Our coverage of Tennessee’s patent suit against Anthropic shows how quickly legal exposure has spread through the sector.

What happens after a demand for information

A civil investigative demand compels documents and answers. Companies typically negotiate scope before producing anything.

Most such investigations close without action. Some end in a consent agreement that changes how a deal is structured. A minority reach court.

Timelines run long. Antitrust reviews of this kind typically run for quarters rather than weeks.

Publicity is its own factor. Companies frequently adjust terms once an investigation becomes public, without ever conceding a problem.

Nothing in the reporting suggests the department has reached a view. It has asked for records, which is where these matters normally begin.

Nvidia has not commented publicly in the reporting reviewed here. Neither has Groq. Both descriptions of the deal so far come from the companies’ own framing at announcement.

That framing deserves scrutiny precisely because it is theirs. Calling a licence non-exclusive tells buyers Groq can still sell to others. It says nothing about whether Groq retains the people needed to do so.

Customers watching the probe face a practical question rather than a legal one. Anyone who designed a deployment around Groq silicon wants to know the roadmap survives the departures.

A Brazilian startup founder running inference at scale cares about supply and price, not about filing thresholds. Regulatory outcomes reach that founder slowly, if at all.

What people are asking about the investigation

  • What is being investigated? Whether Nvidia structured a $20 billion licensing deal with Groq to avoid antitrust review.
  • What was the deal? Groq described it in December as a non-exclusive licence giving Nvidia access to chips optimised for AI inference.
  • Did Nvidia buy Groq? No. Groq remained a separate company, though its chief executive and chief operating officer moved to Nvidia.
  • What step has the department taken? It sent Nvidia a formal demand for information.
  • Does this mean a violation occurred? No. An investigative demand is a request for records, not a finding.
  • Why does AI inference matter here? Inference workloads scale with usage, making inference chips the larger long-term market.

Continue reading on Tamara News

See our reporting on the Nvidia chip export loophole, the memory chip shortage and Tennessee’s patent case against Anthropic.

Sources

  • Bloomberg — DOJ Probes Nvidia’s $20 Billion License Deal With Groq on Antitrust Concerns. bloomberg.com
  • The Daily Guardian — US DOJ probes Nvidia’s licensing deal with AI startup Groq. thedailyguardian.com

Canada Hit the US With $20 Billion in Tariffs — Washington Answered With a Ban

The trade dispute between Washington and Ottawa escalated sharply this month. Canada imposed retaliatory tariffs on roughly $20 billion of US goods. The United States responded with a ban on Canadian alcohol, motorcycles and dairy products. The US Canada trade war has now widened into direct import bans and government contracting restrictions. It deepens a rift between two of the world’s closest trading partners.

US-Canada border crossing point at the center of the US Canada trade war

How the US Canada Trade War Escalated This Month

Canada’s retaliatory tariffs took effect September 8, 2026. They apply to roughly CA$27.6 billion, about US$20 billion, of American goods. Rates range from 15% to 50% depending on the category. The list covers steel, dairy products, household appliances, agricultural equipment and paper products, according to Al Jazeera’s coverage of the tariffs taking effect. The move came in direct response to 50% US tariffs on a similar volume of Canadian goods imposed the previous month. Trade talks between the two governments had broken down before that.

Washington’s Countermove

The United States answered within days. NBC News reported that the US will ban a broad swath of Canadian alcoholic beverages, motorcycles and dairy products from import starting September 29, 2026. President Trump also moved the same week to shut Canadian products out of large, long-term US government contracts. That added a procurement dimension to what began as a tariff dispute.

What This Means for the USMCA

The dispute has widened a rift between two of the closest trading partners in the world. It has cast real doubt on the future of the US-Mexico-Canada Agreement. That deal was meant to keep trade between the three countries largely tariff-free. Industries on both sides of the border, from dairy farmers to motorcycle manufacturers, are now navigating a landscape the USMCA was specifically designed to prevent.

Dairy shipment affected by tariffs in the escalating US Canada trade war

Where the Dispute Goes From Here

The US import ban does not take effect until September 29. That leaves a narrow window for renewed negotiations before the restrictions bite. Both governments face domestic pressure. Canadian exporters are already absorbing the cost of US tariffs imposed last month. American distillers, dairy processors and motorcycle dealers who rely on Canadian imports are bracing for the ban. Analysts say a resumption of formal trade talks is the most likely path to de-escalation. Neither side has signaled a near-term return to the table.

Who Feels the Squeeze First

Dairy farmers on both sides of the border are among the hardest hit, since dairy appears on both countries’ tariff and ban lists. Canadian motorcycle makers face a near-total loss of access to their largest export market once the September 29 ban takes hold. American distillers who rely on Canadian barley and packaging materials are bracing for higher input costs even before any formal ban touches their own products. Steel producers in both countries are watching the broader dispute closely. Steel sits at the center of Canada’s retaliatory list and remains one of the most tariff-sensitive industries in North America. Consumers are likely to notice the impact last, as higher costs typically take a few months to move from producers to store shelves. Prolonged uncertainty makes it harder for manufacturers to plan supply chains that often cross the border multiple times before a finished product reaches a customer, a pressure that tends to push both governments back toward the negotiating table.

Frequently Asked Questions

What did Canada do that triggered this round of the trade war?

Canada imposed retaliatory tariffs of 15% to 50% on about CA$27.6 billion (roughly US$20 billion) of US goods on September 8, 2026. The list covers steel, dairy, household appliances, agricultural equipment and paper products.

How did the United States respond?

The US announced a ban on imports of Canadian alcoholic beverages, motorcycles and dairy products. The ban takes effect September 29, 2026.

Why did Canada retaliate in the first place?

Canada’s move followed 50% US tariffs on roughly $20 billion of Canadian goods imposed the previous month. Several rounds of trade negotiations between the two countries had already collapsed.

Did the US take any other action beyond the import ban?

Yes. President Trump also moved to shut Canadian products out of large, long-term US government contracts on the same day Canada’s retaliatory tariffs took effect.

What does this mean for the USMCA trade agreement?

The breakdown has widened a rift between the two longtime allies. It has cast doubt on the future viability of the US-Mexico-Canada Agreement, which was designed to keep trade between the three countries largely tariff-free.

When do the new US import bans take effect?

The ban on Canadian alcohol, motorcycles and most dairy products is scheduled to begin September 29, 2026.

How quickly the two governments find a path back to the table will determine what comes next. That could be a short-lived standoff, or a longer-term reshaping of North American trade three decades after NAFTA first tied the continent’s economies together. This dispute sits alongside other trade and export fights making headlines this month. See our coverage of the chip export loophole Washington is trying to close and the OPEC+ decision on October oil output.

Sources

153 Million Driver’s Licences Are for Sale. Now We Know Whose.

Identity verification firm IDScan confirmed on 10 September 2026 that attackers took data from its cloud systems. The IDScan data breach involves scans of more than 150 million driver’s licences, alongside millions of other government-issued documents. Reports put the figure at 153 million US and Canadian licences.

The stolen records include full names, driver’s licence numbers and identity numbers taken from documents such as passports. The company is notifying affected people and offering free credit monitoring and identity protection.

The FBI’s New Orleans field office opened an investigation. IDScan operates from Louisiana.

How the IDScan data breach came to light

The public trail started before the company confirmed anything. A dark web platform appeared advertising a searchable database of identity documents for sale.

That listing claimed more than 153 million driver’s licence records, over 10 million ID cards, and millions of further documents including medical cards and travel documents. The FBI began probing the reported breach in early September.

Laptop screen representing exposed records in the IDScan data breach

IDScan said it received information around 1 September indicating that data may have been accessed without authorisation. Its confirmation followed on 10 September.

The breach operators say they pulled data continuously over a year into their private database. That claim comes from the people selling the records, and no independent source confirms it.

The gap between the first public listing and the company’s confirmation ran roughly ten days. Victims therefore learned of the sale before they learned whose systems it came from.

Why an ID verification vendor holds this much

IDScan’s corporate customers range from entertainment venues to cannabis dispensaries. Those businesses check identity documents at the door or at the point of sale.

Verification services sit in the middle of that check. They receive the document image, confirm it, and return a result. Retaining the scan afterwards turns a transaction into a stored record.

That accumulation is what makes vendor breaches different from breaches at any single venue. One compromised supplier exposes every customer that supplier served.

We saw the same structural problem in our coverage of the American Tower data breach and in our report on the TruStage outage lawsuits.

What a stolen licence scan can actually do

A driver’s licence image is more dangerous than a stolen password. Anyone can change a password in seconds. Licence numbers do not work that way.

Document images are the raw material for account takeover at services that verify identity by photo. They also support synthetic identity fraud, where real details from several people are combined into a new profile.

Credit monitoring helps with the visible part. It flags new accounts opened in a person’s name. It does nothing about a document image circulating indefinitely.

Replacing a licence is possible in most jurisdictions but rarely quick. The number often stays the same.

Why this lands differently in each country

The records span both the United States and Canada. Remedies differ across that border and across individual states and provinces.

Most US states let residents freeze credit files with each of the three national bureaus at no cost. Canada operates a smaller bureau market, and the equivalent alert process varies by province.

Replacing a licence also differs. Some jurisdictions issue a fresh number after documented fraud. Others reissue the same number on a new card.

Anyone outside North America is not automatically clear either. The advertised database included travel documents, and passport details travel across borders in a way licence numbers do not.

A researcher from Hanoi who once showed a passport at a venue using this vendor sits in a different position from a local resident. The document she presented carries weight anywhere, and no single agency can reissue confidence in it.

Steps worth taking now

Place a credit freeze rather than relying on monitoring alone. A freeze blocks new accounts instead of reporting them after the fact.

Accept the identity protection if the company notified you. It costs nothing and creates a paper trail if disputes follow.

Treat unexpected identity-verification prompts with suspicion. An attacker holding a document image may attempt verification in your name.

Check statements more often for the next few months. Fraud from bulk data sales often surfaces well after the sale.

Keep the notification letter if one arrives. Banks and insurers frequently ask for proof of the underlying breach before they write off a disputed charge, and a dated letter settles that question faster than a news article does.

Points readers keep raising about the leak

  • How many records were taken? Reports describe scans of more than 153 million US and Canadian driver’s licences, plus over 10 million ID cards and further documents.
  • What data was in them? Full names, driver’s licence numbers and identity numbers from other government-issued documents such as passports.
  • When did IDScan find out? The company said it received information around 1 September and confirmed the breach on 10 September 2026.
  • Is law enforcement involved? Yes. The FBI’s New Orleans field office opened an investigation.
  • What is IDScan offering? Notification of affected individuals plus free credit monitoring and identity protection.
  • How long was data being taken? The breach operators claim exfiltration ran continuously over a year. That claim is theirs and is unverified.

More security reporting on Tamara News

Read our coverage of the American Tower data breach, the Berlin ransomware attack and the TruStage outage lawsuits.

Sources

  • TechCrunch — ID verification giant IDScan confirms data breach with more than 150 million driver’s licenses stolen. techcrunch.com
  • Krebs on Security — FBI Probes Service Selling 153M+ Drivers Licenses. krebsonsecurity.com
  • TIME — FBI Probes Report of Breach Exposing 153 Million Driver’s License Scans. time.com