AI News

Follow the latest AI and automation news

17 Sept 20266 Rabiʻ II 1448 AH

Microsoft AI CEO says AI threats are real, and Anthropic is making it worse

Mustafa Suleyman, CEO of Microsoft AI, discussed the critical need for AI regulation and safety. He recently published a 37-page document titled the "Humanist AI Code of Conduct," outlining the principles guiding AI development. Suleyman also criticized Anthropic’s philosophy regarding AI consciousness, highlighting the potential dangers of this concept in the broader alignment debate.

Arab AI Scene

17 Sept 20266 Rabiʻ II 1448 AH

GISEC 2026 Conference Discusses AI and Cybersecurity

The GISEC 2026 Conference kicked off in Dubai, focusing on artificial intelligence and cybersecurity threats. International experts and officials discussed the impact of emerging technologies and geopolitical changes on the cybersecurity landscape. The sessions highlighted ways to protect critical infrastructure from increasing threats. This event serves as a crucial platform for discussing strategies to enhance cybersecurity amid new challenges, reflecting the importance of AI in shaping future protection strategies.

Business

17 Sept 20266 Rabiʻ II 1448 AH

San Jose Data Center Projects Face Environmental and Economic Challenges

San Jose, the largest city in Silicon Valley, is facing increasing resistance to proposed data center projects as local authorities aim to attract these developments to solidify its position as an AI infrastructure hub. This confrontation highlights the challenges policymakers in the U.S. face in supporting an industry deemed essential for economic competitiveness while addressing rising environmental concerns regarding electricity, water, and pollution.

💡

China Embraces AI with Greater Confidence than the United States

What factors contribute to the greater acceptance of AI in China compared to the U.S.?

Arab AI Scene

Saudi Expertise Takes Center Stage at UNESCO Global Forum in Riyadh

17 Sept 20266 Rabiʻ II 1448 AH

Arab AI Scene

17 Sept 20266 Rabiʻ II 1448 AH

Saudi Expertise Takes Center Stage at UNESCO Global Forum in Riyadh

The world is turning its attention to Riyadh to discuss the future of artificial intelligence (AI) and its ethical implications, as Saudi Arabia continues its efforts to build a generation of national talent specializing in data and AI, placing responsibility and ethics at the core of technological development. These efforts coincide with the fourth UNESCO Global Forum on the Ethics of Artificial Intelligence, where Saudi specialists and researchers are developing AI technologies, focusing on issues such as system reliability, fairness, and addressing challenges related to digital bias and data protection.

Business

How Cooley is accelerating IPO work with ChatGPT

17 Sept 20266 Rabiʻ II 1448 AH

Business

17 Sept 20266 Rabiʻ II 1448 AH

How Cooley is accelerating IPO work with ChatGPT

Cooley has announced the launch of GO Public, a platform utilizing ChatGPT to enhance the initial public offering (IPO) process. The platform aims to assist lawyers in identifying issues earlier and focusing on the most critical judgments. GO Public accelerates legal procedures by leveraging artificial intelligence, allowing lawyers to analyze data more effectively. This innovation contributes to improved efficiency and reduced risks associated with IPOs.

Agents

AI agent swarms are a massive waste of tokens with zero quality gain, says OpenAI Codex developer

17 Sept 20266 Rabiʻ II 1448 AH

Agents

17 Sept 20266 Rabiʻ II 1448 AH

AI agent swarms are a massive waste of tokens with zero quality gain, says OpenAI Codex developer

Eric Provencher, a developer at OpenAI Codex, warns that running more than two parallel sub-agents often leads to unnecessary token consumption without improving quality. He highlighted a project where 1,393 agents spent $20,000 in tokens on a Python refactoring task that a single Astra agent could have completed for a fraction of the cost. This phenomenon is termed the "coordination tax," indicating that agents do not trust each other and end up double-checking each other's work, resulting in resource wastage.

Agents

Adecco Group rolls out Agentforce Coworker to 27,000 staff in 40-plus countries

17 Sept 20266 Rabiʻ II 1448 AH

Agents

17 Sept 20266 Rabiʻ II 1448 AH

Adecco Group rolls out Agentforce Coworker to 27,000 staff in 40-plus countries

The Adecco Group announced the rollout of Salesforce’s Agentforce Coworker in over 40 countries following a successful pilot in the UK and France. This tool supports 27,000 employees in sales and recruitment, facilitating access to data and organizational knowledge. The tool allows staff to access multiple information sources through a single interface, helping them identify potential clients and prepare sales pitches. Recruitment staff can use it to select candidates and initiate screening and onboarding processes, leveraging data from over 2.5 million agent-candidate interactions.

Models

OpenRouter's staggering token chart is the AI bubble debate in a single image

17 Sept 20266 Rabiʻ II 1448 AH

Models

17 Sept 20266 Rabiʻ II 1448 AH

OpenRouter's staggering token chart is the AI bubble debate in a single image

OpenRouter has experienced an astonishing increase in token consumption, surging over 25,000% since January 2025, rising from 0.5 to 126.2 trillion tokens. While this figure appears impressive, it reflects more on token-hungry reasoning models and a growing number of unoptimized AI agents that consume tokens at staggering rates. This trend raises questions about actual AI usage, suggesting a potential bubble in the market, prompting decision-makers to reassess their strategies regarding these models.

Models

Light Origins Open-Sources LightNav-0 Generalist Navigation Brain

17 Sept 20266 Rabiʻ II 1448 AH

Models

17 Sept 20266 Rabiʻ II 1448 AH

Light Origins Open-Sources LightNav-0 Generalist Navigation Brain

Light Origins has announced the launch of LightNav-0, an open-source generalist navigation model built on the Qwen3-VL-4B backbone. The model is released with weights, code, a technical report, and the INSIGHT-Bench evaluation kit, marking an innovative step in navigation technology. The model functions as a navigation brain that shares a single interface across instruction following, open-vocabulary object navigation, and visual tracking, eliminating the need for task-specific prediction heads. Its training process consists of three stages, including online reinforcement learning to improve the policy based on failures.