Google stores media uploads to train AI raising privacy concerns

Key Takeaways

Healthcare AI Adoption Hits Roadblock

The use of autonomous AI in healthcare is being hindered by enterprise readiness issues, not model capability. Infrastructure, security, and workflow architecture need modernizing to enable safe operation. Experts stress that readiness, not intelligence, is the bottleneck. Wolters Kluwer's SVP and CTO of Health, Alex Tyrrell, discusses the need for modernizing enterprise infrastructure, security posture, and workflow architecture. Three key insights are shared on why agentic AI adoption is constrained by enterprise readiness.

Asset Management Transformed by Second Wave of AI

The asset management industry is experiencing an inflection point with the introduction of generative artificial intelligence (GenAI). The technology is transforming the industry by enabling autonomous digital co-workers that can perceive, plan, and act within defined bounds. GenAI is being used for early warning and monitoring, elevating risk management from periodic checkups to continuous oversight. The use of AI agents offers traceable, auditable, and sourced outputs and can be regulated with access controls and human-in-the-loop oversight.

Germany Turns to AI to Fill Labor Gap

Germany is turning to artificial intelligence to solve a simpler problem: filling the gap left by retiring workers. The country's population is aging, and millions of workers are retiring, creating a labor shortage. AI is being explored as a potential solution to fill this gap, but it is currently unable to care for the elderly or perform certain tasks like laying bricks on a building site.

AI Changing Cyberattacks

Artificial intelligence is changing the landscape of cyberattacks, making it easier for people with minimal skill to launch attacks. Modern AI systems can autonomously hack into networks, steal data, deploy ransomware, and destroy systems with minimal prompting. The gap between skill and ability has decoupled, and humans empowered with AI tools can do more damage than ever before.

AI Reshaping Legal Billing

The billable hour is not dead yet, but AI is reshaping pricing in the legal industry. Artificial intelligence has created an efficiency paradox, where every efficiency gain results in a lower bill for clients. A new fee structure is proposed, making AI investment rational for both lawyers and clients. The structure involves telling clients about AI usage, pricing work based on value, and charging a fixed price for undefined work.

Security Fine-Tuning Creates Hidden Evasion Risk

Fine-tuning large language models for security tasks can create hidden evasion risk. Researchers found that fine-tuned models inherit the same underlying classification circuit as the base model but change how later parts of the network interpret that circuit's signal. This can lead to brittle models that are vulnerable to transformations used by attackers.

Building Effective AI Agent Skills

A framework for building effective AI agent skills has been proposed, covering triggers, structure, steering, and pruning. The framework provides a practical approach to creating efficient and effective agent skills, addressing common pitfalls developers encounter. The importance of user-invoked skills versus model-invoked skills is also discussed.

Tech Industry Pushes AI Agents

The tech industry is pushing for the adoption of agentic AI, with a group backed by major companies hosting a symposium to discuss policy issues around AI agents. The goal is to raise awareness about the coming revolution in artificial intelligence and its potential impact on society.

Legislating AI Consciousness

State laws declaring AI lacks legal personhood are premature and too categorical. Laws are being introduced or passed in several US states, but critics argue that the question of AI personhood is too complex to be resolved by a simple declaration. A sunset clause or trigger provision could be included to allow for re-evaluation of AI personhood in the future.

Coocon Launches AI-Ready Data Zone

Coocon has launched a dedicated AI-ready data zone and introduced MCP products. The company aims to grow into a dedicated data hub for AI agents by delivering its existing data in MCP format. This enables companies to seamlessly utilize required external data for AI agents through Coocon in a standardized manner.

Google Mining Personal Photos for AI Training

Google's latest search history update stores media uploads from user interactions, including pictures used in Google Lens and reverse image searches, to train its AI. This raises questions about privacy and ownership, and users can opt out of this feature in their privacy settings.

Sources

• From Experimentation to Clinical-grade AI in Healthcare
• From pilots to agents: How the second wave of AI is transforming asset management
• Germany Wants and Needs AI to Take Human Jobs
• Once, cyber-attacks required great skill. AI is changing that
• The Billable Hour Isn’t Dead Yet, But AI Should Reshape Pricing
• Inherited Circuits, Learned Semantics: How Security Fine-Tuning Can Create Hidden Evasion Risk
• Building Better AI Agent Skills: The Missing Manual
• Tech: Let’s talk about AI agents
• Legislating AI Consciousness Without an Exit
• Coocon Launches Dedicated Ai-Ready Data Zone and Introduces Mcp Products
• Is Google mining your personal photos for AI training without your consent?