AI agents, export curbs and deepfake ads test tech's next phase

Qualcomm is racing ahead with plans for more than 40 new AI-powered devices that could sideline traditional smartphone apps in favor of intelligent agents. CEO Cristiano Amon described these wearables as items people keep on them constantly, from jewelry and camera-equipped earbuds to pins and watches. The goal is constant context about the surrounding world, allowing users to summon an agent for tasks that currently require juggling multiple apps.

Amon emphasized experimentation with form factors during a recent interview. He pointed to agents that handle complex actions like pulling up banking details instantly or booking travel across services. This shift comes as Qualcomm positions itself for consumer electronics that move beyond the phone screen. The company's focus stays on hardware that supports these always-available tools without relying on distant cloud systems.

The push for American AI leadership faces complications from the Trump administration's recent moves. Officials imposed export controls on Anthropic's advanced Mythos 5 and Fable 5 models. Critics argue this step functions like a licensing requirement in disguise, even after earlier decisions scrapped mandatory reporting rules for AI releases. Foreign leaders have taken notice. Canadian Prime Minister Mark Carney warned that overreliance on specific models creates vulnerability, and Europe launched its own tech sovereignty effort to build independent data centers and chip production.

Such interventions send mixed signals about U.S. reliability in AI development. Private companies continue to drive breakthroughs in chips and agents, yet sudden policy shifts risk pushing allies toward alternatives. Qualcomm's work on specialized architectures for smaller gadgets shows how market incentives can accelerate progress without top-down mandates.

Meanwhile, AI tools are reshaping political campaigns through widespread deepfake content. Attack ads now feature fabricated scenes of candidates in compromising situations. In Texas, a Trump-aligned group released a spot showing Democratic Senate nominee James Talarico in a dress performing a song about transgender issues. Other ads have used AI to recreate Talarico reciting old social media posts, with the delivery invented for effect.

Similar tactics appeared in Kentucky's GOP primary, including a deepfake depicting Rep. Thomas Massie with two Democratic congresswomen in awkward social settings. Pro-Massie material featured an elephant wearing a MAGA cap, while opposing spots showed a challenger abandoning a Trump figure in a foxhole. Georgia candidates have produced entirely AI-generated ads. Disclosure of these techniques remains voluntary, leaving voters to sort real footage from generated material.

These developments highlight how AI expands both practical tools and opportunities for distortion. Qualcomm's agent-focused devices promise efficiency gains in daily life, yet the same underlying technology fuels election content that blurs facts. Government efforts to steer AI through export rules add uncertainty to an industry already moving fast on its own. The result is a landscape where innovation from chip designers meets political realities shaped by synthetic media.