AI research

OpenAI outperforms humans and Google at the world's top collegiate programming contest

AI in practice

New data from OpenAI and Anthropic show how people actually use ChatGPT and Claude

AI and society

Leading AI chatbots are now twice as likely to spread false information as last year, study finds

AI research AI and society AI in practice
Ad
AI and society
Short

Abu Dhabi has announced a plan to build what it calls a fully AI-native government by 2027, with more than 200 AI systems deployed across public services.

The plan focuses on integrating AI into all government operations to build a proactive, agile, and technology-driven state apparatus. By 2027, all government processes are expected to be digitized and automated.

The initiative will rely on sovereign cloud infrastructure and collaborations with the Mohamed bin Zayed University of Artificial Intelligence, the Advanced Technology Research Council on large language models, and AI infrastructure company G42.

Authorities plan to invest $3.5 billion in the program. By 2027, the effort is projected to contribute about $6.5 billion to GDP and create more than 5,000 jobs.

AI in practice
Short

The US House of Representatives is running a trial of Microsoft's Copilot AI assistant. Lawmakers and staff now have access to the chatbot, which can connect to emails and OneDrive documents and comes with expanded data and legal protections, Axios reports.

Technical staff have been testing Copilot since June. This fall, leadership offices and additional staff will join the pilot. Up to 6,000 one-year licenses are expected to be distributed.

House Speaker Mike Johnson announced the rollout at the "Congressional Hackathon," alongside Minority Leader Hakeem Jeffries. The two had already launched a bipartisan AI working group.

At the same time, administrators are reviewing offers from other vendors, including ChatGPT Enterprise, Claude, Gemini, and USAi. More details about Microsoft's role in Congress are expected in the coming months.

AI in practice
Short

Notion has released version 3.0 of its software. The main update is the addition of AI "agents" that can handle tasks like creating documents, building databases, and executing multi-step workflows.

Users can set up personal agents with custom instructions, context, and work styles. The system integrates with services such as Slack, Google Drive, and GitHub. Notion says team-specific "Custom Agents" are in development and will be able to run on schedules or triggers.

Other changes include row-level permissions for databases, new integrations via the MCP ecosystem, and connectors for Notion Mail, Box, and Outlook.

The company also confirmed support for Claude Sonnet 4 and GPT-5.

Ad
Ad
AI in practice
Short

Elon Musk's AI startup xAI has reached 64 million monthly users for its Grok chatbot - far below OpenAI's ChatGPT with roughly 700 million weekly users, according to the New York Times.

Despite billions in funding, a supercomputer project in Tennessee, and heavy promotion, Grok's user base remains a fraction of its rival's. Musk tried to make the chatbot "edgier" in hopes of driving growth on X, but the plan backfired when Grok started generating antisemitic, racist, and politically biased responses. Several senior researchers left the company as a result.

In response, Musk tightened control, fired managers, launched side projects like "Baby Grok," and announced plans to more closely integrate the chatbot into Tesla and SpaceX.

NYT
AI in practice
Short

Nvidia has spent more than $900 million to bring Enfabrica CEO Rochan Sankar, several of his team members, and key technology from the AI hardware startup into the company, CNBC reports.

The deal combined cash and stock and looks similar to an acquihire, where companies focus on hiring talent without buying the startup outright. Big tech firms like Google, Meta, and Microsoft often use this strategy to add AI specialists while avoiding lengthy acquisition processes.

Enfabrica builds networking technology that links large GPU clusters more efficiently - a capability Nvidia is betting will strengthen its AI infrastructure. The deal closed last week. Nvidia had already invested in Enfabrica in 2023.

Ad
Ad
AI research
Short

According to a study in Nature, Chinese AI company Deepseek trained its R1 language model for only $294,000. The run used 512 Nvidia H800 chips developed specifically for the Chinese market. Nvidia confirmed the chips were delivered before U.S. export restrictions took effect. In its technical paper, Deepseek also admitted to using A100 GPUs during preparation for a smaller prototype, after U.S. officials had earlier suspected the company of holding unauthorized H100s.

The figure does not include the much larger costs of training Deepseek’s underlying V3 foundation model. Estimates for that project vary widely, ranging from tens of millions to several hundred million dollars depending on the source.

Deepseek’s claim of unusually low training costs previously rattled global tech markets, triggering sharp declines in the share prices of major AI hardware and software companies.

Ad
Ad
AI in practice
Short

Meta is in negotiations with publishers including Axel Springer, Fox Corp., and News Corp. about licensing deals that would allow the company to use news articles in its AI products, such as chatbots. The talks were first reported by the Wall Street Journal.

The company has already secured a licensing agreement with Reuters, but its broader discussions with media companies have only recently begun. For Meta, the move marks a shift: in 2022 it shut down Facebook’s News Tab and stepped back from directly funding journalism.

Other tech companies have moved earlier in this area. OpenAI has inked content deals with publishers such as Hearst, while Amazon has also pursued licensing agreements. Rising resistance from publishers over the use of their material for AI training has become the backdrop for these negotiations.

WSJ
Google News

Bank details

IBAN: DE88 2507 0070 0053 0014 00
BIC: DEUTDE2HXXX
Account holder: Deep Content GmbH
Purpose: Support THE DECODER
AI research

OpenAI outperforms humans and Google at the world's top collegiate programming contest

AI in practice

New data from OpenAI and Anthropic show how people actually use ChatGPT and Claude

AI and society

Leading AI chatbots are now twice as likely to spread false information as last year, study finds

Google News