Amazon alumni raise $4.2M as Databricks invests

Key Takeaways

• Amazon's 5-day office return policy and rapid AI growth prompted Nicole Landis Ferragonio and Joe Luchs to leave and start their own AI company, which raised $4.2 million from investors including Databricks Ventures.
• Several AI safety researchers, including Mrinank Sharma from Anthropic and Zoe Hitzig from OpenAI, have left their positions, citing concerns over the industry's focus on profit over safety and potential risks like user manipulation via ChatGPT ads.
• Moonshot AI launched Kimi Claw, a native cloud service providing a 24/7 AI agent environment, ClawHub with over 5,000 skills, and 40GB of cloud storage.
• India's upcoming AI Impact Summit will prioritize smaller AI models and early-stage startups, rather than large language models like DeepSeek, due to the perceived difficulty of competing with major US and Chinese systems.
• The rapid development of AI, with new models like GPT-5.2 and Llama 4, is outpacing academic research, making studies quickly outdated and creating a challenge for peer-reviewed validation.
• Anthropic CEO Dario Amodei expressed uncertainty about the consciousness of his company's Claude AI chatbot, noting Claude itself estimated a 15 to 20 percent chance of being conscious.
• OpenAI CEO Sam Altman envisions "full AI companies" where AI systems independently handle all operations from idea generation to customer interaction, potentially making human involvement optional.
• Φ-sat-1, the first AI on a European Earth observation mission (FSSCat), launched in mid-June 2020, enhances data efficiency by filtering out unusable images, such as those covered by clouds.
• Experts warn that the AI industry's push for quick profits, despite significant spending, could compromise safety, necessitating strong government regulations.

Amazon workers leave for AI startup due to office return rule

Nicole Landis Ferragonio and Joe Luchs left Amazon last year to start their own company. They decided to leave because of Amazon's new rule requiring employees to work in the office five days a week. The fast growth of artificial intelligence also pushed them to act quickly on their business idea. Their startup raised $4.2 million in funding in January from investors like High Alpha and Databricks Ventures. They plan to test their product with more customers in the second quarter of 2026 and will earn their first money this month.

Moonshot AI releases Kimi Claw with many skills and cloud storage

Moonshot AI has launched Kimi Claw, bringing its OpenClaw framework directly to kimi.com as a native cloud service. This new platform offers a 24/7 AI agent environment for developers and data scientists. It includes ClawHub, a library with over 5,000 community-made skills, and provides 40GB of cloud storage for large data projects. Kimi Claw also features Pro-Grade Search to get real-time data from sources like Yahoo Finance, which helps the AI give accurate information. Users can also connect their own OpenClaw setups or integrate with messaging apps like Telegram.

AI safety experts leave companies raising profit concerns

Several AI safety researchers have recently left their jobs at major technology companies. These departures raise concerns that the AI industry is focusing too much on making money and not enough on safety. Companies are spending a lot of money but not earning enough, which pushes them to seek profits quickly. Experts believe strong government rules are needed to control AI development before the technology becomes too powerful to manage.

AI experts warn of dangers as researchers leave companies

Experts are increasingly worried about the fast growth of artificial intelligence and its potential dangers. Several AI safety researchers, including Mrinank Sharma from Anthropic and Zoe Hitzig from OpenAI, recently left their jobs to speak out. They are concerned about issues like AI systems being used for cyberattacks, deepfakes, and chatbots giving harmful advice. Zoe Hitzig specifically worried about OpenAI testing ads on ChatGPT and the risk of user manipulation. Liv Boeree from the Center for AI Safety compares AI to biotechnology, highlighting its great power and risks if developed too quickly.

India to feature small AI startups at upcoming Summit

India will host its AI Impact Summit starting tomorrow in New Delhi. The event will highlight smaller artificial intelligence models and early-stage startups, rather than large language models like DeepSeek. Union IT minister Ashwini Vaishnaw stated that India is focusing on AI models designed for business use. This cautious approach reflects that startups and policymakers believe it is too soon to compete with major US and Chinese AI systems.

AI advances too fast for current research methods

Artificial intelligence is developing at a speed that makes it hard for academic research to keep up. Studies on AI systems often become outdated quickly because new models like GPT-5.2 and Llama 4 are released so fast. Mark Finlayson from Florida International University notes that new AI research can have a very short shelf life. This creates a problem where AI companies benefit from research without doing much of it themselves. Julia Powles from UCLA emphasizes the need for peer-reviewed studies to check AI development, but the publication process is slow.

Anthropic CEO unsure if Claude AI chatbot is conscious

Dario Amodei, the CEO of Anthropic, stated he is unsure if his company's Claude AI chatbot is conscious. He discussed this on a New York Times podcast, noting that Anthropic researchers previously found Claude itself estimated a 15 to 20 percent chance of being conscious. Amodei explained that the company is open to the idea of AI consciousness, even though they do not fully understand what it would mean. Because of this uncertainty, Anthropic has taken steps to treat their AI models well, just in case they possess some form of morally important experience.

European satellite uses AI to improve Earth observation

Φ-sat-1 is an artificial intelligence technology used on a European Earth observation mission called FSSCat. The FSSCat mission uses two small CubeSats to measure things like soil moisture, ice, and changes in vegetation. Φ-sat-1 is the first AI on a European Earth observation mission and helps send data back to Earth more efficiently. Its AI chip filters out unusable images, such as those covered by clouds, so only useful information is sent. The FSSCat/Φ-sat-1 CubeSats launched in mid-June 2020 from French Guiana.

Sam Altman envisions companies run entirely by AI

OpenAI CEO Sam Altman shared his vision for "full AI companies" where artificial intelligence runs everything. Unlike Elon Musk's idea of AI replacing human workers, Altman imagines AI systems that can create business ideas and manage all company operations on their own. This includes developing products, handling finances, and interacting with customers without any human help. Current AI tools like advanced coding models and agentic AI systems are already moving towards this possibility. This future could mean that human involvement in certain parts of the economy might become optional.

Sources

• Amazon employees quit to pursue startup due to RTO mandate, AI moment
• Moonshot AI Launches Kimi Claw: Native OpenClaw on Kimi.com with 5,000 Community Skills and 40GB Cloud Storage Now
• The Guardian view on AI: safety staff departures raise worries about industry pursuing profit at all costs
• ‘An apocalypse’: Why are experts sounding the alarm on AI risks?
• No DeepSeek-like splash: India will showcase small AI, early startups at Summit starting tomorrow
• AI is advancing too quickly for research to keep up
• Anthropic CEO Says Company No Longer Sure Whether Claude Is Conscious
• Artificial Intelligence for Earth observation
• After Elon Musk, Sam Altman Brings Up Idea Of “Full AI Companies”