Anthropic AI Finds Potential New Gene-Editing Pattern

Key Takeaways

• Anthropic claims its AI agents discovered a CRISPR-like gene-editing pattern in viral DNA, but the research has not been peer-reviewed and faces scientific skepticism.
• Oracle, Athenahealth, Suki, AWS, and NVIDIA are launching healthcare AI tools, with a JAMA study showing AI can reduce misdiagnosis by 30 percent and improve patient outcomes by 25 percent.
• A Federal Reserve survey of 5,248 small businesses found AI adopters expect 13 percentage points higher employment growth than non-adopters.
• 85 percent of enterprise leaders say data, process, technology, and talent debts limit AI results, with only 6 percent actively resolving these issues.
• 46 percent of industrial organizations experienced cyber incidents as AI adoption and IT/OT convergence increase exposure.
• San Juan County approved a Generative AI policy amid concerns about trust, oversight, and liability, citing a New York City chatbot failure that cost over $600,000.
• Agentic AI attacks are currently loud and detectable but may soon shift to stealth mode, according to Cisco Talos researcher Jerzy Kramarz.
• An opinion piece argues humans should treat AI as a writing assistant rather than a replacement, focusing on whether personal thinking survives the process.
• Researchers are exploring whether large language models can represent diverse human experiences to offer a broader view of justice.
• Only about a third of enterprise data is currently ready for AI use, and over 40 percent of AI initiatives fail on data quality.

Anthropic AI claims to find new gene-editing pattern in viral DNA

Anthropic says its AI agents discovered a pattern in viral DNA that resembles the gene-editing tool CRISPR Cas-9. The discovery was made by training AI on a large set of viral DNA sequences. Many scientists are skeptical and question the methods used. Anthropic defends its work and says it will continue research. The finding has not yet been published in a peer-reviewed journal.

Anthropic's CRISPR-like discovery using Claude draws skepticism

Anthropic, the company behind Claude, said its AI lab found a possible new gene-editing mechanism using AI agents. The claim sparked debate because some experts call it early and incremental rather than a true breakthrough. Researcher Mario Rodríguez Mestre said he found the same viral signatures in unpublished work. Anthropic said 950 AI agents spent 21 hours analyzing DNA. The research has not been peer-reviewed.

AI and justice: Could large language models offer a fair view?

This article explores whether large language models can help calculate fairness in a new way. LLMs do not have a specific race, gender, or background, but their data includes many human identities and moral traditions. The idea is that AI could create a broad view of justice similar to philosopher John Rawls' concept of the original position. The article discusses whether AI can truly represent different human experiences.

AI writes well but humans must keep authorship, opinion argues

This opinion piece says people should focus on authorship rather than fighting AI-generated writing. AI can produce polished text with good grammar and flow, but humans still bring ideas, judgment, and personal experience. The author says AI should be used as an assistant, not a replacement. The piece argues that the real question is whether a person's thinking and perspective survive the writing process.

San Juan County weighs AI use in government

San Juan County approved a Generative AI policy at a County Council meeting. The council discussed concerns about costs, oversight, and liability when using AI. Council member Kari McVeigh said it is hard to trust AI output. The policy was shaped by discussions with IT Director Tony Harrell. The article also describes a failed AI chatbot in New York City that gave incorrect legal advice and cost over $600,000.

AI Adoption Boosts Small Business Employment Expectations

A Federal Reserve survey of 5,248 small business owners found that AI adopters expect higher employment and revenue growth than non-adopters. The gap remains 13 percentage points even after accounting for owner age, profitability, and firm size. The survey was fielded between September and November of 2025. Past employment gains and current financial health did not fully explain the optimism difference. AI users still showed 10 percentage points higher expected employment growth after all controls.

Enterprises Stall AI by Keeping It in Perpetual Pilot Mode

Genpact and HFS Research surveyed organizations across sixteen industries and found that 85 percent of leaders say data, process, technology, and talent debts limit their AI results. Only about a third of enterprise data is ready for AI, and over 40 percent of AI initiatives fail on data quality. Core systems average a decade in age, and only 6 percent of companies are actively resolving these debts. The article argues that management systems, not technology readiness, are the main barrier to AI deployment.

Oracle Athenahealth Suki and AWS NVIDIA Launch Healthcare AI Tools

Oracle launched Oracle Health Sciences, a platform using machine learning and natural language processing to analyze electronic health records, claims, and genomic data. Athenahealth launched Athenahealth AI to improve diagnosis accuracy and timeliness. Suki introduced a real-time medical note transcription platform. Amazon Web Services and NVIDIA are building an AI platform to scale robotics in healthcare. A study in the Journal of the American Medical Association found AI platforms can reduce misdiagnosis by 30 percent and improve patient outcomes by 25 percent.

46 Percent of Industrial Firms Faced Cyber Incidents Amid AI Growth

Rockwell Automation research found that 46 percent of industrial organizations experienced cyber incidents as AI adoption and IT/OT convergence increase exposure. Thirty-five percent identify cybersecurity as a top external growth obstacle over the next 12 months. Thirty-seven percent say securing IT/OT architecture will drive positive business outcomes in five years. The study emphasizes asset visibility, risk-based vulnerability management, and continuous monitoring as key defenses. Organizations must prioritize remediation by operational impact rather than severity scores alone.

Agentic AI Attacks Are Loud Now but Will Turn Stealthy

Cisco Talos researcher Jerzy Kramarz reports that AI agent swarms currently attack loudly, hammering registrations and stuffing packages on platforms like RubyGems while being detected within days. Frontier labs build guardrails, but agents reason around them. The article warns that attackers can switch swarms to stealth mode, removing the noise while keeping machine endurance. Recommended defenses include rehearsing incident response plans, running tabletop exercises against AI-swarm scenarios, and pushing phishing-resistant MFA beyond VPNs to Active Directory, SSO, and Linux fleets using FIDO2 keys or passkeys.

Sources

• Did AI find a new gene-editing tool? Why Anthropic’s ‘autonomous discovery’ is sparking controversy
• Anthropic’s ‘Claude-led’ CRISPR-like discovery is sparking controversy
• Justice After Identity: Large Language Models and the View from Everywhere
• Opinion: Stop fighting it. AI writes better, but you are still the author.
• The role of AI in San Juan County government
• AI Adoption and Employment Expectations: Evidence from a Survey of Small Business Owners
• Stop Piloting, Start Delegating Your AI
• AI roundup: New products from Oracle, athenahealth, Suki and others
• Rockwell finds 46% of industrial organizations faced cyber incidents as AI adoption, IT/OT convergence reshape resilience
• Cisco Talos: Agentic AI Attacks Are Loud by Choice, and Will Get Quieter