Microsoft Challenges OpenAI with Homegrown Models as Zuckerberg Predicts Billions of AI Agents
Microsoft pitched its own homegrown AI models and harnesses to Wall Street on Wednesday, openly competing with OpenAI and Anthropic more than ever before, according to Reuters. The company highlighted a Mythos competitor and outlined plans for continued growth across its 1 million plus enterprise customers, signaling a strategic shift away from sole reliance on OpenAI's GPT-4 technology.
Microsoft CEO Satya Nadella confirmed during an earnings call on Wednesday that a Copilot "super app" combining chat, coding, and agentic capabilities will launch in 2025, reports The Verge. Nadella said the app will span both consumer and commercial experiences, describing the evolution from chat to Cowork to Autopilots as Copilot reaches over 40 million monthly active users.
Meta CEO Mark Zuckerberg predicted that billions of people will have personal AI agents within 5 years, according to CNBC. As Meta pours tens of billions of dollars into AI infrastructure and agent development, Zuckerberg is working to convince investors that the long-term payoff across its 3.2 billion daily active users will justify the massive capital expenditure.
A government ban on foreign-made robots may hinder instead of help the US robotics industry, reports Wired. The policy could cut off access to critical components and supply chains that American robotics firms rely on, potentially slowing domestic innovation in a market projected to reach $74 billion by 2030.
An AI avatar of Brazil's jailed former president Jair Bolsonaro, who is banned from communicating publicly, is testing new AI rules ahead of Brazil's 2026 elections, according to the Financial Times. Lawyers for the former president insist he did not authorise the projection following a legal challenge from leftwing groups, raising novel questions about AI-generated political speech in the country's 2026 electoral cycle.
A Mythos attack on a 3rd-round post-quantum cryptography algorithm candidate has put it out of commission, according to The Register. The HAWK algorithm withstood years of testing that had yet to uncover a fatal weakness, but the Mythos method exposed vulnerabilities in under 72 hours, highlighting emerging risks for AI-assisted cryptanalysis.
Sources: