AI IMPLICATIONS

Trump’s latest defense: Blame it on AI

President Donald Trump is blaming AI for a viral video showing trash bags being tossed out of a White House window, even after his own press team offered a conflicting explanation. 

The clip posted on X shows trash bags flying from a top-floor window inside the president’s residence. Shared by thousands, the video itself has been viewed more than 10 million times.

At a Tuesday press briefing in the Oval Office, Trump dismissed the video, saying “that’s probably AI-generated.”

“Actually, you can’t open the windows. You know why? Because they are all heavily armored and bulletproof.” 

Trump went on to say the first lady had complained that she couldn’t let fresh air in because the windows are sealed, bulletproof, and each weighs about 600 pounds, so “you have to be pretty strong to open them up.” 

Watching the video from a reporter’s phone, Trump seized the moment to make a wider point about the problems with AI.  

Trump’s latest AI defense fits a familiar pattern of the president using technology as a scapegoat during viral moments or political controversy.

In August 2024, during his presidential campaign, Trump falsely accused Kamala Harris of using AI to inflate the size of the crowd at her rally in Detroit earlier that month. While running for the GOP nomination in 2023, he took to the platform accusing his opponents of using AI “in their fake television commercials” to make him look “pathetic.” 

Yet it’s not all about accusations. In May, Trump signed the Take It Down Act into law, criminalizing the publishing of revenge porn, including AI-generated deepfakes, a key initiative by the first lady to protect victims of digital exploitation.

TOGETHER WITH IBM

New report reveals insights on AI-driven attacks

Our study found AI-driven attacks accounted for 1 in 6 data breaches.

Attackers can use generative AI to perfect and scale their phishing campaigns and other social engineering attacks. Gen AI can reduce the time needed to craft a convincing phishing email from 16 hours down to only five minutes on average.

Learn more

LEGAL BATTLES

Meta-backed Scale AI sues rival, former employee

Meta-backed Scale AI sues rival, former employee 

Meta-backed Scale AI is suing rival data-labeling startup Mercor and a former employee, Eugene Ling, for allegedly stealing trade secrets and breach of contract in a case that underscores the escalating battle over trade secrets in the AI space. 

The lawsuit, filed Wednesday in a San Francisco federal court, addresses Mercor’s theft of Scale AI’s trade secrets and proprietary business information. 

“As a small and new competitor vying for a share of the market, Mercor sought to bypass the time and investment required to develop its own business strategies by illicitly acquiring those of Scale, the industry leader,” the suit states. 

It goes on to say that in the weeks leading up to his departure, Ling stole more than 100 confidential documents that outlined Scale’s customer strategies and other proprietary information. 

According to the lawsuit, a significant portion of the stolen documents relate to a client Mercor failed to win over as its own after specifically recruiting Ling to expand its relationship with that customer. The suit further claims that Mercor recruited Ling specifically to gain access to Scale’s client strategies, giving the company an unfair advantage in the competitive GenAI market.

The lawsuit also alleges that: 

• Ling stole confidential documents while still employed at Scale, including materials related to that customer and other clients, as well as proprietary rubrics and off-the-shelf data packs

• While exploring new opportunities, Ling transferred these files from Scale’s Google Drive to his personal drive using multiple methods

• Mercor’s leadership, including its CEO and president, allegedly coordinated with Ling and continues to rely on him on that customer account despite knowledge of the stolen materials

In a statement to Bloomberg, Mercor co-founder Surya Midha said that while the company has hired several former Scale employees, “we have no interest in any of Scale’s trade secrets and in fact are intentionally running our business in a different way.” Midha added that Mercor never accessed the documents Ling claimed to have and that the company is now conducting an internal investigation.

TOGETHER WITH ZED

Claude Code, Now in Zed

You can now run Claude Code directly inside Zed, with multibuffer reviews, cross-file edits, and full syntax awareness.

It’s the control and context you need, without leaving your editor. And since it’s powered by Zed's Agent Client Protocol, you can use Claude Code alongside Gemini CLI—or bring your own agent.

Try Claude Code in Zed

AGENTIC AI

New AI agent set to take on OpenAI, Anthropic

DeepL is changing its game, moving beyond translation and writing tools with an AI agent designed to automate everyday business tasks. 

The company announced Wednesday the release of its first general-purpose AI assistant equipped with the company’s “precise understanding of language and context,” which DeepL said will change the game when it comes to how businesses get work done.

Currently valued at $2 billion, the German company launched in 2017 and was solely focused on AI-powered translation services, until now. 

The launch signals DeepL’s biggest shift yet, bringing the trusted translation service into the crowded enterprise AI space, positioning it against big players including Microsoft, OpenAI, Anthropic and Salesforce. 

“DeepL Agent can understand your instructions in almost any language, just as easily as a co-worker can,” the company posted on X. “It can tackle almost any task you would use a computer for, and it’s able to integrate with your existing tools, systems and workflows so that it can start assisting you straight away. “

DeepL is already an expert in solving complex language problems, a focus the company said gives the new agent an edge when it comes to reasoning and collaboration. 

The company said that, unlike other AI agents, the DeepL Agent was specifically developed for business use and can take on almost any computer tasks from navigating systems and websites to planning work, solving problems and even making and executing decisions.

LINKS

• OpenAI increases size of its secondary share sale by more than $4 billion

• Robotics maker Unitree set to file Q4 IPO, downplays concerns

• Tesla’s new master plan puts robotics and AI front and center 

• Anguilla hits the jackpot with its national .ai website domain

• AI is reshaping packaging robotics boosting factory productivity

• HappyRobot raises $44 million to expand AI agents for freight operations

• Microsoft cracks two practical problems with analog optical computer

• Nvidia acquires three-year-old software development startup Solver

• ElevenLabs SFX v2: Generate endless looping sound effects with a single prompt

• HappyRobot: Build fully custom AI workers that handle tasks across the organization

• Ada: Create professional reports from any dataset

• Sidekick: Build Zapier-style automations using only a chat interface

• Luminate: Senior Data Scientist, Generative AI

• Venmo: Data Scientist

• OpenAI: Data Lead, Robotics

• Hunter Bond: AI/ML Engineer

• HartleyCo: Data Scientist (Growth)

GAMES

A QUICK POLL BEFORE YOU GO