AP report: Hegseth warns Anthropic to let the military use company's AI tech as it sees fit

Defense Secretary Pete Hegseth issued Anthropic CEO Dario Amodei a Friday deadline to allow unrestricted military use of the company's AI technology or risk losing its Pentagon contract. The ultimatum applies to a $200 million defense contract awarded last summer alongside Google, OpenAI, and xAI. Pentagon officials threatened to designate Anthropic a supply chain risk or invoke the Defense Production Act to compel compliance. The pressure campaign follows months of tension between Anthropic's state-level AI regulation advocacy and the administration's deregulation agenda.

Breaking encryption with a quantum computer just got 10 times easier

Researchers have developed a more efficient quantum computing approach using qLDPC codes that reduces qubit requirements for breaking RSA encryption by an order of magnitude. The new method enables qubits to interact beyond nearest neighbors, increasing information density and reducing the estimated qubit count from millions to approximately 100,000.

Facial recognition error prompts police to arrest Asian man for burglary 100 miles away

Thames Valley Police arrested Alvi Choudhury after automated facial recognition software falsely matched him with footage of a burglary suspect in Milton Keynes, 100 miles away. The match was made despite the suspect appearing "10 years younger" with clear physical differences. Choudhury, who was held for 17 hours before being cleared, is claiming damages and calling for transparency about wrongful arrests involving facial recognition technology. The case adds to documented concerns about demographic bias in automated facial recognition systems.

Ministers urged to impose temporary ban on crypto political donations

UK Parliament's Joint Committee on National Security Strategy demanded a moratorium on cryptocurrency political donations until safeguards against foreign interference are implemented. The warning comes after the Representation of the People bill omitted crypto donation restrictions despite concerns that digital assets enable malign actors to conceal funding sources, complicating Electoral Commission and law enforcement oversight capabilities.

CrowdStrike says attackers are moving through networks in under 30 minutes

CrowdStrike's 2026 Global Threat Report found the average time from intrusion to lateral network movement plummeted to 29 minutes in 2025, a 65% increase in attack speed. State-sponsored threat actors increased cloud intrusion activity by 266% while AI-enabled adversary activity surged 89%. Chinese threat groups achieved immediate system access in two-thirds of vulnerability exploitations, with 40% targeting edge devices. North Korea's Lazarus Group orchestrated the largest cryptocurrency theft in history stealing $1.46 billion from Bybit.

600+ FortiGate Devices Hacked by AI-Armed Amateur

A Russian-speaking threat actor used commercial generative AI tools to compromise more than 600 Fortinet FortiGate firewalls across 55 countries in five weeks. Researchers from Amazon Web Services found the attacker was not state-sponsored but used LLMs to automate attack scripting, credential extraction, and lateral movement. The campaign exploited exposed management ports and weak credentials, with AI enabling the low-skilled actor to achieve outcomes previously requiring substantial technical expertise.

US antitrust enforcers to revamp guidelines on rivals collaborating

The Department of Justice and Federal Trade Commission launched a public inquiry to develop updated antitrust guidance for businesses on competitor collaborations, including data sharing and pricing information exchange. The agencies seek input by April 24 on how new technologies have changed competitive dynamics, building on 2000-era guidelines that predate modern data aggregation and AI-driven business alliances.

I.R.S. Tactics Against Meta Open a New Front in the Corporate Tax Fight

The Internal Revenue Service is using real-world profit data to challenge how Meta and other large technology companies value intellectual property moved offshore, opening a new front in the government's battle against corporate tax avoidance. The agency is scrutinizing the "Double Irish" arrangement and transfer pricing mechanisms that allowed Meta to relocate IP rights to Ireland while its U.S. parent maintained control of core technologies, questioning whether offshore subsidiaries paid adequate consideration for assets that generate billions in global revenue.

Will Trump's DOJ actually take on Ticketmaster?

Gail Slater, who led the Justice Department's Antitrust Division, departed just weeks before the agency is set to face trial against Live Nation, creating uncertainty around federal enforcement in one of the year's most significant antitrust cases. At least 40 states joined the DOJ lawsuit, and state attorneys general have signaled they will continue the litigation regardless of federal participation, with California's top antitrust enforcer Paula Blizzard affirming trial will proceed March 2.

Employers are winning the gig worker messaging war

A poll shows 76% of Americans support keeping app-based workers as independent contractors if employers provide portable benefits. The Trump administration is rolling back a Biden-era labor rule that had steered companies toward classifying workers as employees. Labor advocates argue the independent contractor model inhibits union formation.

SDA taps AST SpaceMobile to demo commercial satellite links to military radios

The Space Development Agency awarded AST SpaceMobile USA a $30 million contract under the Hybrid Acquisition for proliferated Low-earth Orbit program's Europa Track 2 initiative. The company will use its BlueBird satellite constellation—currently six satellites in orbit with a seventh launching soon—to demonstrate direct tactical communications with existing military radios. Unlike traditional proprietary military satellite systems, AST's "bent-pipe" architecture uses commercial infrastructure to provide high-bandwidth data transport from low Earth orbit for defense applications.

Hundreds of FortiGate Firewalls Hacked in AI-Powered Attacks: AWS

Amazon Web Services threat researchers identified a Russian-speaking hacker who compromised over 600 Fortinet FortiGate firewall instances across 55 countries using generative AI tools. The attacker exploited exposed management ports and weak credentials, then used AI to generate Python scripts for credential extraction and lateral movement. AWS confirmed the threat actor is not associated with any advanced persistent threat group, demonstrating how commercial AI services lower technical barriers for unsophisticated attackers to execute scaled campaigns.

The FDA creates a quicker path for gene therapies

The FDA announced draft guidance for a "plausible mechanism pathway" that would allow gene-editing treatments for ultra-rare diseases to proceed without traditional clinical trials when biological understanding is strong. The policy builds on the successful treatment of "Baby KJ," an infant who received a bespoke CRISPR therapy for a fatal metabolic disorder. The framework specifically targets diseases too rare to attract pharmaceutical investment, potentially opening treatment paths for thousands of conditions affecting 30 million Americans.

The Big One: The cyberattack scenarios that keep officials up at night

Seven former national security officials and industry leaders detailed their gravest cybersecurity concerns. Paul Nakasone, former NSA and Cyber Command head, warned that nation-state actors who have breached food and water infrastructure could accidentally trigger catastrophic outages if they lose control of AI agents. Former CISA director Jen Easterly noted AI is scaling existing weaknesses in insecure software and over-trusted automation.

Building the backbone for Europe's biodiversity monitoring

Researchers from University of Amsterdam and German Centre for Integrative Biodiversity Research (iDiv) have published a roadmap in Nature Reviews Biodiversity proposing a European Biodiversity Observation Network (BON) to unify fragmented national monitoring systems. The plan combines environmental DNA sampling, satellite remote sensing, and citizen science data through a proposed European Biodiversity Observation Coordination Centre (EBOCC). The initiative aims to standardize data workflows around Essential Biodiversity Variables to support the Kunming-Montreal Global Biodiversity Framework and European Green Deal commitments.

Voters know what the next big issue is. They don't know how they feel about it.

New polling by POLITICO reveals data centers are emerging as a potent but undefined political issue ahead of 2026 elections, with Democrats seeing traction from candidates who campaigned on regulating data center energy consumption and water usage. Republican Oklahoma City Mayor David Holt noted data centers shifted from unknown to omnipresent in political discourse within months. The survey found only 25% of Americans believe data centers won't play a role in their local elections, while bipartisan momentum builds to accelerate permitting and energy infrastructure to support AI growth despite environmentalist warnings of conflict with net-zero goals.

Is Age Verification a Trap?

Age verification mandates force platforms to store biometric data, ID images, and verification logs for regulatory defense, creating persistent privacy risks. Facial age estimation systems produce false positives that lock accounts for days while platforms must retain sensitive data long enough to prove compliance decisions to regulators, transforming child safety infrastructure into permanent identity surveillance architecture.

Live facial recognition to be used ahead of Everton v Man Utd

Democrats and Republicans agree on one thing: regulating the use of AI

Bipartisan consensus has emerged across US state legislatures regarding the need for AI and data center regulation despite historical partisan gridlock on technology policy. Republican and Democratic lawmakers at state capitols nationwide are advancing regulatory frameworks addressing artificial intelligence deployment and data center infrastructure expansion. The rare bipartisan agreement signals growing institutional recognition that algorithmic systems have outpaced existing regulatory frameworks governing their deployment and social impact.

Password Managers Share a Hidden Weakness

Researchers at ETH Zurich and USI Lugano have exposed fundamental flaws in password manager cryptographic implementations, challenging "zero knowledge" claims that companies claim prevent them from accessing user credentials. The study demonstrates that malicious insiders or sophisticated hackers can exploit these cryptographic weaknesses to compromise the supposedly secure vaults across multiple major platforms. The findings undermine years of privacy assurances that have positioned password managers as essential security infrastructure.

Democrats oppose Trump administration's tech buildup for immigration enforcement

Democratic lawmakers and civil liberties advocates are pushing back against the Trump administration's deployment of facial recognition and biometric surveillance technologies to support deportation operations. Representative Pramila Jayapal has introduced the ICE Out of Our Faces Act to ban ICE and Customs and Border Protection from acquiring and using biometric identification systems, while requiring deletion of existing data. Civil rights advocates note facial recognition systems have documented accuracy disparities for women and people of color, raising concerns about wrongful arrests.

SerpApi asks court to dismiss Google web scraping lawsuit

A federal judge has allowed SerpApi's antitrust counterclaims against Google to proceed, finding the scraping company plausibly alleged that Google holds monopoly power in the search market and that its lawsuit against SerpApi could constitute exclusionary conduct. The ruling challenges the limits of how a monopolist can use Terms of Service to restrict access to information that has become essential internet infrastructure.

America desperately needs new privacy laws

Congress has repeatedly failed to pass comprehensive privacy legislation despite decades of corporate surveillance expansion. The Verge reports that even targeted measures like the Fourth Amendment Is Not for Sale Act—which would restrict police from using data brokers to bypass privacy laws—have stalled. Tech monopolies exacerbate privacy problems by reducing competition and centralizing information in exploitable silos, while new technologies from AR glasses to generative AI create fresh surveillance risks faster than regulatory frameworks can adapt.

Have we leapt into commercial genetic testing without understanding it?

Daphne O. Martschenko and Sam Trejo's new book "What We Inherit" warns that polygenic embryo selection has entered clinical practice with minimal regulatory oversight while offering limited predictive value. The technology uses statistical associations between gene variants and traits to rank embryos, but accuracy varies dramatically by genetic ancestry — with Pacific Islander Americans seeing systematically worse predictions than those of European descent. If access remains concentrated among wealthy populations, embryo selection could encode class and racial disparities directly into the human genome, compounding across generations.

How A.I. Money Is Flooding Into the Midterm Elections

AI companies and allied groups have spent at least $83 million on federal elections and are positioned to deploy hundreds of millions more for the 2026 midterm elections in an extraordinary demonstration of political power from Silicon Valley. OpenAI, Anthropic, and affiliated executives are directing record political contributions as the industry moves to shape favorable regulatory frameworks and secure government contracts. The spending surge comes as AI companies face mounting regulatory scrutiny and competition for lucrative defense and infrastructure contracts. Some donors have broader political interests beyond AI policy, but the coordinated industry investment represents an unprecedented electoral mobilization by emerging technology firms.

Top NATO allies believe cyberattacks on hospitals are an act of war. They're still struggling to fight back.

A major poll across the US and four NATO member states reveals that majorities in each country view cyberattacks on critical infrastructure, particularly hospitals and energy grids, as acts of war. Despite this consensus, the allied nations remain divided on appropriate responses, with less than half believing that hacking political leaders' private communications constitutes an act of war. State-linked attacks have escalated dramatically - the 2024 Change Healthcare breach exposed 190 million US medical records, while a Russian cyberattack on UK NHS systems contributed to a patient's death. Iranian government-backed hackers also targeted Boston Children's Hospital in 2022.