Anthropic refuses Pentagon AI limits as Google launches Gemini features

Key Takeaways

• Anthropic refused the Pentagon's request to remove safety limits on its AI, citing ethical concerns regarding potential use in mass surveillance and autonomous weapons.
• The University of Kentucky launched a new AI literacy course for staff, part of its CATS AI strategy, focusing on responsible AI use and societal impact.
• Google's March Pixel Drop introduced new AI features like Circle to Search, virtual 'Try It On,' and Gemini's in-app task capabilities.
• MEXC rolled out a six-tool AI trading suite to over 1.5 million users, including AI News Radar and AI Trending Search, enhancing investment decisions.
• 98% of US product leaders expect generative AI to improve internal workflows within three years, anticipating better decision-making and data security, with varying preferences for OpenAI, Google, and Microsoft.
• Researchers discovered over 200 'AI slop' scam websites, created using basic AI prompts, demonstrating how AI lowers the cost and time for scaling fraudulent activities.
• LivePerson launched Syntrix, a platform for testing and managing customer-facing AI agents, aiming to reduce operational costs by up to 30% in agent training and mitigate risks.
• Lenovo's CIO advises patience with AI investments, stressing the need for AI to prove its economic value and ROI despite rapid learning capabilities.
• Evo 2, an open-source AI, was developed and trained on 8.8 trillion DNA bases to identify complex genomic features, with precautions taken to prevent misuse.

University of Kentucky launches new AI course for staff

The University of Kentucky's AI Literacy and Training (ALT) Hub has created a new online course for faculty and staff. This training is part of the Commonwealth AI Transdisciplinary Strategy (CATS AI) to promote responsible AI use. The course has six modules covering AI basics, prompting, stress-testing ideas, fact-checking, personal responsibility, and societal impact. It takes about four hours to complete core lectures, or ten hours with the final project. The goal is to make AI accessible and useful for everyone at the university.

UK ALT Hub offers new AI literacy training for employees

The University of Kentucky's AI Literacy and Training (ALT) Hub has launched a new online course for its faculty and staff. This initiative is part of the university's CATS AI strategy, which focuses on the responsible use of artificial intelligence in various sectors. The course, adapted from an undergraduate curriculum, aims to introduce AI concepts and practical applications. It includes six modules covering AI vocabulary, prompting, stress-testing ideas, fact-checking, responsible use, and societal impacts. Participants can complete the training at their own pace, with an option to engage in a final project.

Anthropic and Pentagon clash over AI ethics and governance

AI company Anthropic refused the Pentagon's request to remove safety limits on its AI technology, leading to a standoff. The Department of Defense wanted to use Anthropic's AI for mass surveillance and autonomous weapons, but Anthropic declined, citing ethical concerns. This conflict highlights the lack of a clear legal framework for AI governance in the U.S. The situation raises questions about who should control AI development and deployment, especially when ethical lines conflict with national security demands. Congress is urged to establish laws to address these complex issues.

Google Pixel adds new AI features in March update

Google's March Pixel Drop introduces new AI-powered features for personalization and convenience. Users can now use Circle to Search to identify items in images and virtually 'Try It On' for clothing. Gemini can now perform tasks within apps, and Magic Cue helps find restaurant recommendations in chats. The update also includes improvements to the Now Playing app, custom icons, security features, and Pixel Watch capabilities. Some AI features like Scam Detection are also expanding to new regions.

MEXC launches AI trading tools, reaching 1.5 million users

MEXC has completed the rollout of its six-tool AI trading suite, serving over 1.5 million users. These tools, including AI News Radar and AI Trending Search, work together to help investors make decisions. The suite aims to provide institutional-grade trading infrastructure to all users. MEXC's Chief Operating Officer stated that AI is now fundamental to trading, enhancing research and decision-making. The company notes that a significant percentage of Gen Z traders already use or trust AI in their investments.

US product leaders see AI boosting workflows

A new report shows that 98% of US product leaders expect generative AI to improve internal workflows within three years, a significant increase from previous years. They also anticipate better decision-making accuracy and stronger data security. While confidence in AI's benefits is high, there are varying preferences for AI providers like OpenAI, Google, and Microsoft. The report also highlights that product leaders are increasingly viewing AI as a tool to enhance oversight, compliance, and customer experience, moving beyond just efficiency gains.

Fraudsters use AI to create over 200 scam websites

Researchers have discovered a network of more than 200 'AI slop' websites created by a single group using basic AI prompts. The operators accidentally exposed their prompts in the website code, revealing how AI is used to quickly generate content for scams. These websites aim to trick users into sharing sensitive data or to earn advertising revenue from duped advertisers. The operation, named 'AutoBait' by cybersecurity firm DoubleVerify, shows how AI significantly lowers the cost and time needed to scale such fraudulent activities.

LivePerson launches Syntrix for AI agent training

LivePerson has launched Syntrix, a new platform designed to help businesses confidently deploy and manage customer-facing AI agents. Syntrix allows companies to test AI agents and train live agents using simulated customer scenarios before real interactions. This helps address concerns about AI hallucinations and brand compliance. The platform aims to accelerate AI deployment, reduce operational costs by up to 30% in agent training, and mitigate risks by identifying issues in a simulated environment.

Lenovo CIO advises patience with AI investments

Lenovo's Global Chief Information Officer, Arthur Hu, believes patience is key in AI investing, even as the technology matures. He notes that companies are willing to invest in generative AI for rapid learning, but ROI is still a consideration. Hu emphasizes that AI integration is a gradual process, starting with smaller use cases and expanding across the enterprise. He also addresses employee concerns by framing AI as a tool to enhance productivity and reshape workloads rather than replace jobs. While boards are pushing for faster deployment, Hu stresses the need for AI investments to eventually prove their economic value.

New open source AI trained on trillions of DNA bases

Researchers have developed Evo 2, an open-source AI trained on trillions of DNA bases from all three domains of life: bacteria, archaea, and eukaryotes. This advanced AI can identify key features in complex genomes, such as regulatory DNA and splice sites, which are often difficult for humans to spot. Evo 2 was trained on the OpenGenome2 dataset, containing 8.8 trillion bases. The researchers excluded viruses that infect eukaryotes to prevent potential misuse. This development marks a significant step in interpreting complex genomic structures.

Sources

• UK ALT Hub launches new AI literacy course for faculty and staff
• UK ALT Hub launches new AI literacy course for faculty and staff
• The Conscience Clause: Anthropic-Pentagon Standoff Reveals Who Governs AI
• March Pixel Drop: New personalization and AI tools
• MEXC Completes Rollout of Six-Tool AI Trading Suite, Reaching Over 1.5 Million Users
• 98% of US Product Leaders Expect Gen AI to Improve Workflows
• Exclusive: Fraudsters create 200+ AI slop websites in one operation
• Syntrix Launches as the First AI Agent Evaluation and Training for CX Teams
• Lenovo’s CIO says patience is a virtue in AI investing, but the clock is ticking
• Large genome model: Open source AI trained on trillions of bases