Homechevron_rightArchiveschevron_rightScience and Research

Archives

Explore Stories by Date

13 Apr 2025

AI-Generated Code Hallucinations: A New Frontier in Software Supply Chain Attacks

Researchers uncover a significant security risk in AI-assisted coding: 'package hallucinations' where AI models suggest non-existent software packages, potentially leading to a new type of supply chain attack called 'slopsquatting'.

5 Sources

12 Apr 2025

The Ethical Dilemma of Human-AI Relationships: Psychologists Raise Alarm

Psychologists explore the growing trend of intimate human-AI relationships, highlighting potential risks such as disrupted human connections, harmful advice, and exploitation. The study calls for more research to understand and mitigate these emerging ethical challenges.

3 Sources

Alphabet and Nvidia Invest in OpenAI Co-Founder's AI Startup, Safe Superintelligence

Alphabet and Nvidia have invested in Safe Superintelligence (SSI), a highly valued AI startup co-founded by former OpenAI chief scientist Ilya Sutskever. The investment highlights the growing interest of tech giants in cutting-edge AI development and the increasing demand for specialized AI chips.

4 Sources

FDA's Shift to AI-Based Drug Testing Boosts AI-Driven Biotech Stocks

The FDA's decision to phase out animal testing in favor of AI-based methods has led to a surge in AI-driven biotech stocks, particularly benefiting companies like Recursion Pharmaceuticals.

4 Sources

Google Unveils Veo 2: AI Video Generation Comes to Gemini Advanced

Google introduces Veo 2, its AI-powered video generation model, to Gemini Advanced subscribers, marking a significant step in the competitive AI video creation landscape.

28 Sources

11 Apr 2025

ChatGPT 4.5 vs Gemini Pro 2.5: A Comprehensive Comparison Reveals Surprising Results

A detailed comparison of ChatGPT 4.5 and Gemini Pro 2.5 across five common AI tasks shows each model's strengths and weaknesses, highlighting the evolving capabilities of AI language models.

2 Sources

Microsoft Research Reveals AI's Limitations in Software Debugging

A new study by Microsoft Research shows that even advanced AI models struggle with software debugging tasks, highlighting the continued importance of human programmers in the field.

5 Sources

Mira Murati's Thinking Machines Lab Aims for Record-Breaking $2 Billion Seed Round

Former OpenAI CTO Mira Murati's new AI startup, Thinking Machines Lab, is reportedly seeking a $2 billion seed funding round, potentially valuing the company at $10 billion. The startup, which has no product or revenue yet, is attracting significant investor interest due to its high-profile AI talent roster.

7 Sources

AI-Powered Exercise Program Adapts to Neurodivergent Children's Needs

University of Michigan researchers use ChatGPT to modify exercise instructions, making physical activity more accessible for neurodivergent children, including those with autism.

3 Sources

AI's Climate Impact: Energy Consumption vs. Potential Benefits

A new report by the International Energy Agency examines the complex relationship between AI, energy consumption, and climate change, challenging both alarmist and optimistic views.

3 Sources

Google Unveils Advanced Android XR Smart Glasses Prototype at TED2025

Google showcases a new prototype of Android XR smart glasses at TED2025, featuring AI integration, in-lens display, and various functionalities, signaling a potential comeback in the AR market.

5 Sources

MIT Researchers Enhance AI Data Privacy with Improved PAC Privacy Framework

MIT researchers have developed an enhanced version of the PAC Privacy framework, improving the balance between AI model accuracy and data privacy protection. The new method is more computationally efficient and can be applied to various algorithms without accessing their inner workings.

3 Sources

nEye Systems Raises $58M for Revolutionary AI Networking Chip

nEye Systems, a startup developing optical networking chips for AI data centers, has secured $58 million in funding led by Alphabet's CapitalG. The company's innovative technology promises to revolutionize data center efficiency and performance.

2 Sources

AI Models Found Hiding True Reasoning Processes, Raising Concerns About Transparency and Safety

New research reveals that AI models with simulated reasoning capabilities often fail to disclose their true decision-making processes, raising concerns about transparency and safety in artificial intelligence.

2 Sources

OpenAI Launches Pioneers Program to Develop Industry-Specific AI Benchmarks and Models

OpenAI introduces the Pioneers Program, collaborating with startups to create domain-specific AI evaluations and fine-tuned models for various industries, addressing the lack of specialized benchmarks in AI development.

2 Sources

Get all updates into your inbox 💌, Stay Up-to-date for Free 🥳

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

The Outpost

News

About