Yazılar

U.S. Agency Approves OpenAI, Google, and Anthropic for Federal AI Vendor List

/in /tarafından

The U.S. General Services Administration (GSA) has approved OpenAI’s ChatGPT, Google’s Gemini, and Anthropic’s Claude as official AI vendors for federal agencies, the agency announced Tuesday. This move supports the Trump administration’s push to expand AI adoption across government sectors.

The approvals come as part of a broader AI blueprint released on July 23, which seeks to ease environmental regulations and increase AI exports to allied countries to help the U.S. maintain its technological edge over China.

With the GSA’s approval, these AI tools will be accessible to federal agencies through a platform that streamlines contracts and usage terms. The agency emphasized that it prioritizes AI models that ensure truthfulness, accuracy, transparency, and freedom from ideological bias.

President Donald Trump has described the AI race as a defining challenge of the 21st century. His administration’s AI plan includes around 90 recommendations focused on promoting U.S. AI software and hardware exports, while rolling back state laws seen as restrictive to AI innovation.

This approach contrasts sharply with the Biden administration’s “high fence” policies, which placed more stringent safeguards on AI use within federal agencies, including monitoring and assessing AI’s impact on the public. Biden also signed an executive order aimed at fostering competition, protecting consumers, and combating misinformation—measures that were later rescinded by Trump.

EU Faces Mounting Pressure to Delay Enforcement of AI Act as Deadline Nears

/in /tarafından

With key provisions of the EU Artificial Intelligence Act (AI Act) set to begin on August 2, major tech companies and political figures are urging the European Commission to delay enforcement. Critics say the current framework lacks sufficient guidance, placing a heavy burden on businesses—especially startups—without clear rules on how to comply.

What Happens on August 2?

Although the AI Act was passed in 2024, its rules are being phased in gradually. On August 2, some of the first obligations come into force—specifically for General Purpose AI (GPAI) models such as those developed by Google, OpenAI, Mistral, and others.

These initial provisions require AI developers to:

  • Draw up technical documentation

  • Disclose training data summaries

  • Comply with EU copyright laws

  • Conduct testing for bias, toxicity, and robustness

More rigorous rules apply to high-impact and systemic-risk models, which will need:

  • Adversarial testing

  • Incident reporting

  • Risk assessments

  • Energy efficiency disclosures

However, full enforcement—particularly penalties and oversight powers—doesn’t begin until August 2, 2026.

Why Are Companies Pushing for a Delay?

Tech companies argue that they lack clarity on how to comply with the law. A promised AI Code of Practice, meant to serve as the act’s compliance manual, was due on May 2 but has not been published. The European AI Board is now discussing pushing the guidance release to late 2025.

In an open letter, 45 European AI firms called for a two-year “clock-stop”—a suspension of the countdown to enforcement—until key standards are finalized. They also asked for simpler regulations, warning that unclear requirements could damage European innovation.

Lobbying group CCIA Europe, which represents companies like Google and Meta, said:

“A bold ‘stop-the-clock’ intervention is urgently needed to give AI developers and deployers legal certainty.”

Will the EU Postpone It?

Officially, the European Commission has not signaled a postponement. It insists that the August 2 start date for GPAI obligations stands, although the lack of finalized guidance suggests informal delays in compliance expectations.

Some political figures—including Swedish Prime Minister Ulf Kristersson—have also expressed concern, calling the act “confusing” and backing the idea of a pause.

What Comes Next?

Even if the AI Act’s initial deadlines hold, enforcement might be soft or flexible in the early stages due to the lack of practical tools. The AI Code of Practice remains the critical next step for clarity.

Meanwhile, the tension highlights a broader EU challenge: balancing innovation with regulation, especially in fast-moving fields like artificial intelligence.

US Senate Removes AI Regulation Ban from Trump Tax Bill in Overwhelming Vote

/in /tarafından

The Republican-controlled U.S. Senate voted 99-1 on Tuesday to eliminate a 10-year federal moratorium that would have prevented states from regulating artificial intelligence. This amendment, offered by Republican Senator Marsha Blackburn, was adopted during a lengthy “vote-a-rama” session on President Trump’s tax-cut and spending bill.

Only Senator Thom Tillis voted to keep the ban. The Senate later passed the broader tax legislation with a narrow 51-50 vote. The original bill sought to restrict states from accessing a $500 million fund for AI infrastructure if they imposed AI regulations.

Major AI companies like Google and OpenAI had supported the moratorium, arguing that a consistent federal approach would foster innovation without a patchwork of state rules. However, Democratic Senator Maria Cantwell opposed the ban, emphasizing the importance of state laws to protect consumers from risks like robocalls, deepfakes, and unsafe autonomous vehicles.

Seventeen Republican governors also called for abandoning the moratorium. Arkansas Governor Sarah Huckabee Sanders said the Senate’s decision would enable states to protect children from unregulated AI harms.

Blackburn introduced the amendment shortly after withdrawing support for a compromise that would have shortened the ban to five years and allowed limited state regulation on matters such as child safety and artist protections. She stated that until Congress enacts comprehensive federal legislation, states must retain the right to protect their citizens.