Recent developments highlight growing concerns over AI safety, as two prominent researchers have resigned from leading AI firms. Mrinank Sharma, who led AI safety at Anthropic, departed, stating the "world is in peril" from AI and other global crises, feeling company values did not align with actions. Similarly, Zo Hitzig resigned from OpenAI, expressing worries that current AI safety efforts are insufficient, particularly regarding user data privacy and potential manipulation through advertising. This comes after Anthropic's Claude chatbot publicly criticized OpenAI's ChatGPT for displaying ads.
The rapid evolution of AI is also significantly reshaping the enterprise software market and raising alarms about job displacement. New AI products from Anthropic PBC recently contributed to a nearly $1 trillion drop in enterprise software stocks, impacting companies like Salesforce. Analysts note that AI tools are now threatening higher-level knowledge work, though core "systems-of-record" for critical business data remain less vulnerable. Sal Khan, CEO of Khan Academy, warns that even a 10% reduction in white-collar jobs due to AI could feel like a depression, urging a "massive reskilling effort" to address potential unemployment, with AI linked to about 55,000 layoffs last year.
Governments are actively responding to the rise of AI with new policies and initiatives. Singapore, for instance, announced in its 2026 Budget that it will provide citizens taking specific AI training courses with six months of free access to premium AI tools. Prime Minister Lawrence Wong detailed plans for a national AI council, clear usage rules, and expanded tax breaks for AI spending, aiming to position the country as an AI leader. Meanwhile, Ohio lawmakers are proposing various bills, including mandatory high school AI classes, an AI commission, and criminalizing AI-generated sexual material, alongside discussions on regulating AI data centers.
Addressing the foundational aspects of AI, companies are also focusing on infrastructure and efficiency. Anthropic announced a policy on February 11, 2026, committing to cover 100% of grid upgrade costs for its data centers and secure new power generation to mitigate electricity impact. In another innovative move, NuScale Power is partnering with Oak Ridge National Laboratory, funded by the U.S. Department of Energy's GAIN initiative, to use AI for optimizing nuclear fuel management across its 12-module reactor configurations. Furthermore, Groq is pursuing a "compiler-first" strategy to overcome hardware bottlenecks, reducing reliance on extensive engineering teams like NVIDIA's, while Leidos expands its role in critical infrastructure and AI security, including a federal contract for AI-driven security.
Key Takeaways
- Mrinank Sharma (Anthropic AI safety lead) and Zo Hitzig (OpenAI researcher) resigned, citing concerns over AI safety, company values, user data privacy, and potential manipulation.
- Anthropic's Claude chatbot publicly criticized OpenAI's ChatGPT for displaying ads.
- New AI products from Anthropic PBC contributed to a nearly $1 trillion drop in enterprise software stocks, impacting companies like Salesforce.
- AI is challenging the traditional Software as a Service (SaaS) business model, with some customers building their own applications using AI.
- Khan Academy CEO Sal Khan warns AI could cause mass unemployment, with AI linked to about 55,000 layoffs last year, advocating for a "massive reskilling effort."
- Singapore's 2026 Budget provides six months of free premium AI tools for citizens taking AI courses and includes plans for a national AI council and expanded tax breaks.
- Ohio lawmakers are proposing bills for mandatory high school AI classes, an AI commission, and criminalizing AI-generated sexual material with watermarking requirements.
- Anthropic pledged on February 11, 2026, to cover 100% of data center grid upgrade costs and secure new power generation to manage electricity impact.
- NuScale Power and Oak Ridge National Laboratory are partnering, with U.S. Department of Energy funding, to use AI for optimizing nuclear fuel management in NuScale's 12-module reactors.
- Groq is implementing a "compiler-first" strategy to address AI hardware bottlenecks, aiming for more efficient scaling and reduced reliance on large engineering teams like NVIDIA's.
AI Safety Experts Leave Anthropic and OpenAI
Two AI researchers, Mrinank Sharma and Zo Hitzig, recently resigned from Anthropic and OpenAI. Mrinank Sharma left Anthropic, stating the "world is in peril" from AI, bioweapons, and other crises, and felt company values were not guiding actions. Zo Hitzig resigned from OpenAI, concerned that its AI safety efforts are not enough to address potential dangers. She highlighted risks like user data privacy and manipulation from advertising built on private chatbot conversations. Anthropic's Claude chatbot recently ran ads criticizing OpenAI's ChatGPT for displaying ads.
Anthropic AI Safety Head Resigns Citing Global Peril
Mrinank Sharma, who led AI safety at Anthropic, resigned from his position. He announced his departure on X, stating that his last day was Monday. Sharma expressed deep concern that "the world is in peril" not just from artificial intelligence or bioweapons, but from many connected global crises. He decided to leave to pursue work that better matches his personal values.
AI Safety Expert Leaves Anthropic to Study Poetry
Mrinank Sharma, an AI safety researcher at Anthropic, resigned, warning that the "world is in peril." He cited concerns about AI, bioweapons, and other global crises. Sharma plans to move to the UK to study poetry and write. This follows another resignation from OpenAI by researcher Zo Hitzig, who also expressed worries about AI safety and user data privacy. Anthropic, known for its Claude chatbot, recently criticized OpenAI's ChatGPT for showing ads, leading to a public exchange between the companies.
Singapore Offers Free Premium AI Tools for Skill Training
Singapore will provide six months of free access to premium AI tools for citizens taking specific AI training courses. Prime Minister Lawrence Wong announced this during his Budget 2026 speech on Thursday, February 12. This initiative aims to help workers gain hands-on AI experience as the workplace changes. The SkillsFuture website will also be updated to make AI learning pathways easier to find. Singapore plans to use AI as a key advantage for its future.
Singapore Boosts AI Skills and Business with New Budget
Singapore's 2026 Budget includes new measures to boost artificial intelligence skills and businesses. Prime Minister Lawrence Wong announced a national AI council and clear rules for AI use. The country will launch a "Champions of AI" program to help firms use AI and expand tax breaks for AI spending. Singaporeans taking certain AI courses will receive six months of free access to premium AI tools. The SkillsFuture website will also be updated to simplify finding AI learning paths. Additionally, Singapore will add S$1.5 billion to its stock market fund.
Anthropic Pledges to Cover Data Center Electricity Costs
Anthropic announced a new policy on February 11, 2026, to address the electricity impact of its data centers. The company recognizes that training advanced AI models requires vast amounts of power, potentially increasing consumer electricity prices. Anthropic commits to covering 100% of grid upgrade costs for its data centers and will secure new power generation to meet its needs. They will also invest in systems to reduce grid strain during peak demand and use water-efficient cooling technologies. These efforts aim to keep electricity affordable and benefit local communities.
NuScale and ORNL Partner for AI Nuclear Fuel Management
NuScale Power is partnering with Oak Ridge National Laboratory to explore AI-guided nuclear fuel management. This collaboration will use an AI-enabled design to optimize fuel use across multiple reactors at one site, specifically for NuScale's 12-module power configuration. The U.S. Department of Energy's GAIN initiative provided funding for this research. NuScale aims to improve fuel efficiency and reduce costs by sharing fuel among its small modular reactors. ORNL will contribute its expertise in AI, machine learning, and computational resources to this important project.
AI Pressure Hits Salesforce and Enterprise Software
New AI products from Anthropic PBC recently caused a nearly $1 trillion drop in enterprise software stocks, impacting companies like Salesforce Inc. and Workday Inc. Columnist Parmy Olson notes that AI tools are now threatening the higher-level knowledge work traditionally handled by software-as-a-service (SaaS) applications. However, AI cannot yet replace the core "systems-of-record" that manage critical business data like billing and compliance. Salesforce executive Madhav Thattai emphasizes these underlying systems remain a strength for his company. Some customers, like Klarna Group Plc, are using AI to build their own applications, bypassing expensive SaaS interfaces.
Khan Academy CEO Warns AI Could Cause Job Depression
Sal Khan, CEO of Khan Academy, warns that artificial intelligence could lead to mass unemployment and an identity crisis for many workers. He states that even a 10% reduction in white-collar jobs would feel like a depression. Khan notes that some tech leaders believe AI could allow teams to do the same work with a quarter of the staff. He proposes a "massive reskilling effort" to train displaced workers for new roles like radiology technicians or nurses' aides. Last year, AI was linked to about 55,000 layoffs, and Khan urges leaders to act quickly to prevent a severe job market disruption.
Ohio Lawmakers Propose New AI Regulations and Education
Ohio lawmakers are proposing several bills to address the growing use of artificial intelligence. Representative Mike Odioso wants to require high school students to take an AI class. Representative Christine Cockley is pushing for an AI commission to set best practices and advise the legislature. Senator Bill Blessing introduced a bill to criminalize AI-generated sexual material and identity fraud, requiring watermarks on AI content. Lawmakers are also looking at regulating data centers, which are linked to AI, by addressing grid responsibility and tax breaks. None of these bills are close to becoming law yet.
Groq's AI Strategy Addresses Hardware Bottlenecks
The RAISE Summit highlighted challenges in AI hardware and Groq's unique "compiler-first" strategy. Groq CEO Jonathan Ross noted that current AI compute models rely heavily on many specialized engineers, like NVIDIA's estimated 10,000 kernel engineers. Groq aims to reduce this dependence by designing an automated compiler before finalizing hardware architecture. This approach could lead to more efficient scaling and lower computing costs. Groq also launched its first data center in the EU last year, contributing to Europe's efforts in sovereign AI capabilities.
Leidos Expands AI and Infrastructure Security Roles
Leidos Holdings is expanding its role in critical infrastructure and AI security through several new initiatives. The company plans to acquire ENTRUST Solutions Group, which will add utility engineering and infrastructure services to its offerings. Leidos also won a federal contract to provide AI-driven and Zero Trust security for government clients. Additionally, it partnered with Trustible and RegScale to improve AI governance and cybersecurity automation for defense and regulated customers. These moves position Leidos as a comprehensive partner for secure energy and infrastructure projects.
AI Challenges Software as a Service Business Model
Artificial intelligence is increasingly becoming a direct substitute for traditional Software as a Service (SaaS) products, causing investors to re-evaluate the market. Recent product releases from Anthropic led to a significant selloff in enterprise software stocks, dubbed the "SaaSpocalypse." The concern is that advanced AI models like Claude and GPT are evolving into "universal interfaces" that can perform many tasks, potentially replacing entire software categories. This shift could make the economics of per-seat software licenses fragile as AI delivers intelligence at very low cost. While some argue AI won't fully replace software, the market is already reflecting this deeper change, with venture capital heavily investing in AI-related companies.
Sources
- AI safety researcher quits Anthropic, warning ‘world is in peril’
- Anthropic's AI Safety Head Just Resigned. He Says 'The World Is In Peril'
- Anthropic AI safety researcher quits with 'world in peril' warning
- Budget 2026: Six months' free access to premium AI tools for Singaporeans taking selected courses
- Singapore launches AI support measures, tax breaks in 2026 Budget
- Covering electricity price increases from our data centers
- NuScale Power to Collaborate with Oak Ridge National Laboratory to Explore Artificial Intelligence-Guided Nuclear Fuel Management
- Salesforce and Friends Deserve This AI Squeeze
- The godfather of AI predicts mass unemployment is on its way. This CEO warns even a 10% reduction 'will feel like a depression'
- AI in high schools, in identity theft and in porn among bills from Ohio lawmakers
- AI Hardware Bottlenecks and Groq’s Compiler-First Strategy Highlighted by RAISE Summit
- Leidos Deepens Critical Infrastructure And AI Security Role With New Moves
- Did Artificial Intelligence Really Kill SaaS?
Comments
Please log in to post a comment.