Congress Extends Surveillance Law, White House Eyes AI Model Rules

Congress Extends Surveillance Law Hours Before Expiration

  • Congress approved a 45-day extension for Section 702 of FISA.
  • The program permits warrantless surveillance of foreign targets.
  • The bill advances to President Donald J. Trump for signature.

WASHINGTON D.C., May 04 (TNGB) – Lawmakers moved quickly to avoid any gap in key intelligence authorities under Section 702 of the Foreign Intelligence Surveillance Act. The House passed the clean 45-day extension by a 261-111 vote while the Senate approved it by unanimous consent just hours before the previous 10-day patch expired at midnight April 30. The program allows collection of communications involving non-U.S. persons located abroad without individual warrants, though it has faced scrutiny over incidental U.S. person data handling. The legislation now proceeds to President Donald J. Trump for signature.

This extension gives Congress until mid-June to negotiate longer-term reforms.

Why This Matters: Short-term continuity preserves intelligence capabilities while debates over warrants and oversight continue in a divided Congress.

White House Considers Tighter Rules for New AI Models

  • Officials weigh pre-release review requirements for advanced AI systems.
  • Concerns over Anthropic Mythos capabilities prompted the policy shift.
  • An executive order could establish a formal government vetting process.

WASHINGTON D.C., May 04 (TNGB) – The Trump administration is examining options for greater oversight of powerful new AI models before they reach the public. Discussions include creation of a working group with technology executives and potential mandatory safety or security reviews for frontier systems. The move follows internal unease about rapid capability advances and expansion plans for models such as Anthropic’s Mythos, which has demonstrated strong performance in identifying software vulnerabilities. White House officials have briefed companies including Anthropic, Google, and OpenAI on the proposals under consideration.

The policy reversal signals a more hands-on federal approach to frontier AI development.

Why This Matters: Pre-release scrutiny could slow deployment of advanced models while aiming to reduce national security and misuse risks.

Threat Groups Target Critical Infrastructure for Data Theft

  • Two Com-affiliated clusters conduct rapid SaaS-focused extortion campaigns.
  • Voice phishing and SSO abuse enable quick data exfiltration without malware.
  • Targets span academic, aviation, financial, and technology sectors.

USA, May 04 (TNGB) – CrowdStrike detailed the activities of Cordial Spider and Snarky Spider, two financially motivated threat clusters tied to the broader Com ecosystem. Active since at least October 2025, the groups rely on vishing and adversary-in-the-middle pages mimicking single sign-on portals to compromise identity systems and move laterally through cloud environments. They prioritize speed, often beginning data theft within an hour of initial access, and demand seven-figure ransoms while threatening public leaks. The operators focus on U.S. organizations in critical infrastructure and related industries.

Defenders must prioritize identity security and SaaS visibility to counter these fast-moving actors.

Why This Matters: High-speed, low-malware extortion campaigns raise the stakes for organizations protecting sensitive data in cloud platforms.

South Africa Withdraws AI Policy Draft Over Fake Citations

  • Officials pulled the draft national AI policy after reference errors surfaced.
  • At least six of 67 citations referenced nonexistent academic papers.
  • AI tools generated the fabricated sources without human verification.

SOUTH AFRICA, May 04 (TNGB) – The Department of Communications and Digital Technologies withdrew its first Draft National Artificial Intelligence Policy after journalists identified fictitious references. The document, published for public comment in April, contained citations to journals and papers that do not exist, which investigators attributed to unverified use of generative AI during drafting. Communications Minister Solly Malatsi described the lapse as unacceptable and announced that consequence management would follow for those responsible. The department will revise the policy before reissuing it for comment.

The incident exposed gaps in quality control when governments adopt AI assistance for complex regulatory work.

Why This Matters: Reliance on unverified AI output can undermine the credibility of national policy documents and delay governance frameworks.

Sony PlayStation Settlement Offers Small Consumer Payouts

  • A $7.85 million class-action settlement covers digital game purchases.
  • Eligible users bought titles on PlayStation Network from 2019 through 2023.
  • Automatic account credits are expected after final court approval in October.

USA, May 04 (TNGB) – Sony Interactive Entertainment reached a proposed $7.85 million settlement in a long-running antitrust class action alleging inflated prices for digital games on the PlayStation Store. The agreement covers U.S. users who purchased more than 100 specific titles directly through PlayStation Network between April 1, 2019, and December 31, 2023. A federal judge in California granted preliminary approval, with a fairness hearing scheduled for October 15, 2026. If finalized, funds will be distributed as account credits rather than cash, after attorneys’ fees, across potentially millions of eligible profiles.

Individual payouts are expected to remain modest given the large number of potential claimants.

Why This Matters: Automatic small credits provide limited relief to consumers while resolving claims without admission of liability by Sony.

Influential ChatGPT Education Study Retracted Over Flaws

  • A meta-analysis touting educational benefits of ChatGPT faced retraction.
  • Independent review identified discrepancies in data analysis and methodology.
  • The journal removed the paper from the scientific record.

USA, May 04 (TNGB) – Researchers had published a meta-analysis highlighting positive effects of ChatGPT and similar large language models on student learning outcomes across multiple studies. Subsequent examination revealed inconsistencies in how results were aggregated and reported, prompting the journal to retract the influential paper. The retraction followed standard post-publication review processes that flagged methodological concerns. Academic observers noted the case illustrates broader challenges in rapidly evolving research on generative AI applications in education.

The withdrawal removes one prominent data point from ongoing debates about classroom AI integration.

Why This Matters: Retractions of high-profile AI education studies underscore the need for rigorous verification before policymakers or educators adopt new tools.

Lawmakers Question Federal Defense of Data Centers

  • A House hearing examined growth in data centers and rising adversary threats.
  • Lawmakers debated whether existing critical infrastructure authorities suffice.
  • Expansion of facilities has increased their strategic importance and exposure.

WASHINGTON D.C., May 04 (TNGB) – Members of Congress reviewed the rapid buildout of data centers and the corresponding increase in targeting by foreign adversaries seeking espionage or disruption opportunities. Witnesses highlighted gaps in coordinated federal protection for the sector, which supports cloud computing, AI training, and essential digital services. The discussion centered on whether data centers warrant formal designation and resources comparable to other critical infrastructure sectors such as energy and finance. No immediate legislative action was announced.

The hearing reflects growing recognition that digital infrastructure now sits at the center of national security planning.

Why This Matters: Stronger federal coordination could reduce vulnerabilities in the backbone of the modern economy and government operations.

Pixel 11 Faces Potential Impact from RAM Shortage

  • Industry-wide memory constraints may affect upcoming smartphone specifications.
  • Google could reduce RAM allocations in the next Pixel series to manage costs.
  • AI-driven demand continues to strain global semiconductor supply chains.

USA, May 04 (TNGB) – Persistent shortages of dynamic random-access memory components are influencing design decisions across the smartphone industry. Manufacturers face higher costs and allocation limits as data center and AI accelerator production consumes significant capacity. Google’s Pixel team is reportedly evaluating configurations that balance performance, price, and availability for the Pixel 11 family expected later this year. Similar pressures have already appeared in other premium Android and iOS device roadmaps.

Consumers may see incremental rather than dramatic memory upgrades in mid-cycle flagships.

Why This Matters: Component shortages can slow feature improvements and keep device prices elevated for everyday users.

Australia Teen Social Media Ban Sees Widespread Evasion

  • The under-16 social media ban took effect in December 2025.
  • Surveys indicate more than 60 percent of teens still maintain access.
  • Many users rely on parental assistance or technical workarounds.

AUSTRALIA, May 04 (TNGB) – Four months after Prime Minister Anthony Albanese’s government implemented minimum-age requirements for major platforms, compliance remains uneven. Age-verification systems have blocked or removed millions of accounts, yet independent surveys show a majority of 12- to 15-year-olds continue to use at least one restricted service. Common bypass methods include parental identification documents and simple profile alterations. Platform companies have removed preexisting underage accounts in large numbers, but new or evasive access persists.

Enforcement difficulties reveal limits of technical age gates when user incentives and family dynamics favor continued access.

Why This Matters: Persistent teen usage despite legal restrictions tests the practical limits of age-based platform regulation.

  • The Windows editor’s maintainer objected to an unauthorized macOS version.
  • Trademark concerns centered on potential user confusion and brand dilution.
  • The independent developer agreed to rename the port Nextpad++.

USA, May 04 (TNGB) – Don Ho, creator of the popular open-source Notepad++ text editor, issued a formal demand regarding an unofficial macOS port that used the original name and branding. He cited trademark protections and the risk of confusion among users who expect consistent behavior across platforms. The Mac developer responded by committing to the new name Nextpad++ for future releases and removing the contested branding. The exchange occurred publicly through project channels and developer communications.

Open-source maintainers increasingly assert trademark rights to preserve project identity across unofficial ports.

Why This Matters: Clear naming boundaries help users distinguish official tools from independent adaptations in crowded software ecosystems.