Meta CyberSOCEval, Google AI Data Cleaning, Amazon Agent Hiring

The artificial intelligence landscape is rapidly evolving, with significant developments across cybersecurity, healthcare, and enterprise workflows. CrowdStrike and Meta have launched CyberSOCEval, an open-source benchmark designed to test and evaluate the performance of AI models, particularly large language models, in cybersecurity operations like incident response and threat analysis. This initiative aims to help security teams select effective AI tools to combat increasingly sophisticated cyber threats. In healthcare, URAC has introduced the first AI accreditation program to establish national standards for ethical and safe AI implementation in patient care, focusing on transparency, bias management, and patient safety. This follows calls for responsible innovation and clear standards in the sector, with new regulations emerging, such as California's law requiring human oversight for AI denying healthcare coverage. Ambient-listening AI tools are also emerging to reduce clinician burnout by automating note-taking. Meanwhile, enterprise AI is seeing advancements in agentic capabilities. Soul Machines has integrated its Digital Workers with ServiceNow's AI Platform via the Workforce Connect App, enabling emotionally intelligent AI agents for customer and employee interactions. Google DeepMind is addressing the challenge of data scarcity by developing methods to clean and reuse 'toxic' data, such as biased or misinformation-laden content, for AI training, potentially extending the usability of existing data resources. Amazon is bolstering its AI agent development by hiring senior executives for its AWS division to lead foundational infrastructure and development environments for agentic AI. However, the rise of autonomous AI agents also presents oversight challenges for publishers and advertisers, with platforms like Salesforce and Adobe introducing tools that make automated decisions. Protocols for AI agent interactions are being developed, but concerns remain about data trails and accountability. In a legal development, Apple faces a class-action lawsuit from authors alleging the company used pirated books from the Books3 dataset to train its AI models, seeking damages and an injunction. Public relations professionals are also adapting, needing strong story mining skills to ensure brand visibility in an AI-driven media environment. Separately, in naval maintenance, Global Spatial Technology Solutions is using its AI system, OCIANA Predictive Maintenance, to forecast equipment failures for the Canadian Navy. Marketers are also gaining new tools, with Ubersuggest adding AI features to help track and improve brand visibility on conversational AI search engines.

Key Takeaways

• CrowdStrike and Meta have partnered to release CyberSOCEval, an open-source benchmark for evaluating AI performance in cybersecurity operations.
• URAC has launched the first Health Care Artificial Intelligence (AI) Accreditation Program to set national standards for ethical AI use in patient care.
• Soul Machines has integrated its Digital Workers with ServiceNow's AI Platform to enhance enterprise workflows with emotionally intelligent AI agents.
• Google DeepMind has developed a method to clean and reuse 'toxic' data for AI training, addressing data scarcity concerns.
• Amazon is expanding its focus on AI agents by hiring senior executives for its AWS division to lead agent infrastructure and development.
• The increasing autonomy of AI agents is creating new oversight and accountability challenges for publishers and advertisers, with platforms like Salesforce and Adobe introducing agentic AI tools.
• Authors have filed a class-action lawsuit against Apple, alleging the company used copyrighted books without permission to train its AI models.
• Public relations professionals need to develop strong story mining skills to ensure brand visibility in the AI era, leveraging AI tools for trend identification.
• Global Spatial Technology Solutions is using its AI-driven OCIANA Predictive Maintenance system to forecast equipment failures for the Canadian Navy.
• Ubersuggest has introduced new AI Search Optimization features to help marketers track and improve brand visibility on conversational AI search engines.

CrowdStrike and Meta create AI benchmarks for cybersecurity

CrowdStrike and Meta have partnered to release CyberSOCEval, a new set of open source benchmarks. These benchmarks will help organizations test and evaluate how well AI systems perform in security operations. CyberSOCEval uses Meta's CyberSecEval framework and CrowdStrike's threat intelligence to assess AI models on tasks like incident response and threat analysis. This aims to help security teams choose effective AI tools to combat growing cyber threats.

CrowdStrike, Meta launch new AI tool for security

CrowdStrike and Meta have teamed up to launch CyberSOCEval, a new set of open source benchmarks. This tool is designed to measure how well artificial intelligence, specifically large language models, perform in high-pressure cybersecurity situations. The benchmarks are built on Meta's CyberSecEval framework and CrowdStrike's threat intelligence. CyberSOCEval will help security teams evaluate AI models for tasks like incident response and threat analysis, ensuring better defense against advanced attacks.

CrowdStrike and Meta simplify AI security tool evaluation

CrowdStrike and Meta have launched CyberSOCEval, a new set of benchmarks to help businesses evaluate AI security tools. The suite tests large language models (LLMs) on key cybersecurity tasks like incident response and threat analysis. This aims to help organizations choose the best AI-powered tools for their needs. The benchmarks are now available to the AI and security community to improve model capabilities and combat AI-driven cyber threats.

Healthcare AI needs responsible innovation and clear standards

The healthcare industry must develop responsible AI practices to maximize its potential while ensuring patient safety and data privacy. Key best practices include reducing bias, protecting patient data, and continuous monitoring of AI outputs. Ambient-listening AI tools are helping reduce clinician burnout by automating note-taking. Organizations like the Coalition for Health AI are promoting transparency in AI use. New regulations, like California's law against AI denying healthcare coverage, aim to ensure human oversight.

URAC launches first AI accreditation for healthcare

URAC has launched the Health Care Artificial Intelligence (AI) Accreditation Program to set national standards for ethical AI use in patient care. This program provides a framework for safe, ethical, and equitable AI implementation. It includes tracks for AI developers and user organizations, focusing on transparency, bias management, and patient safety. The accreditation aims to build trust and ensure AI fulfills its promise in healthcare without compromising patient well-being.

Soul Machines app humanizes AI workflows on ServiceNow

Soul Machines has launched the Workforce Connect App on the ServiceNow AI Platform, integrating its Digital Workers with enterprise workflows. This app allows businesses to deploy emotionally intelligent Digital Workers for customer and employee-facing tasks. By using ServiceNow's Now Assist, the integration aims to provide empathetic and efficient service. This collaboration enhances productivity by enabling AI agents to automate tasks and collaborate across systems.

Google DeepMind cleans toxic data for AI training

Google DeepMind has developed a method to clean and reuse 'toxic' data, such as content with bias or misinformation, for AI training. This innovation addresses the growing scarcity of high-quality training data, which is projected to be exhausted by 2026. The data detoxification process removes harmful elements while preserving informational value. This approach could help AI models continue to advance responsibly and extend the usability of existing data resources.

AI agents create oversight challenges for publishers and advertisers

The rise of autonomous AI agents that instruct other agents is creating new oversight and accountability challenges for publishers and advertisers. Platforms like Salesforce and Adobe are introducing agentic AI tools that make decisions and adapt automatically. Protocols are being developed for AI agents to interact online, but this raises concerns about data trails and responsibility if mistakes occur. Businesses need to establish governance frameworks to manage these AI systems effectively and protect consumers.

Amazon hires executives to boost AI agent development

Amazon is strengthening its focus on AI agents by hiring two senior executives for its AWS division. David Richardson returns to lead AgentCore, the company's foundational agent infrastructure. Joe Hellerstein, a renowned database researcher, joins to advance Kiro, AWS's integrated development environment for agents. These hires signal Amazon's aggressive move to become a leader in agentic AI, where software agents build and manage complex applications for users.

PR pros need story mining skills for the AI era

In the age of AI, public relations professionals must develop strong story mining skills to ensure brand visibility and credibility. AI-powered search increasingly relies on unpaid and earned media content. Securing media mentions is challenging due to faster news cycles. PR pros need business fluency, a newsroom mentality, and close ties to business operations to uncover and frame compelling narratives. AI tools can support this process by identifying trends and tailoring content.

AI system improves naval maintenance in Canada

Global Spatial Technology Solutions (GSTS) is advancing its OCIANA Predictive Maintenance (OPM) system under a Canadian government contract. This AI-driven tool forecasts equipment and system failures in naval fleet operations. By analyzing real-time vessel, environmental, and operational data, the OPM system provides situational awareness and predictive insights. This supports proactive maintenance planning for maritime operators.

Ubersuggest adds AI features for marketers

Ubersuggest has launched new AI Search Optimization features to help marketers track and improve their visibility on conversational AI search engines. These tools provide insights into brand AI visibility, consumer search behavior, and competitor performance. The platform integrates these AI features with existing SEO tools like rank tracking and content generation. Ubersuggest aims to make enterprise-level AI intelligence accessible and affordable for marketers.

Apple sued for using books in AI training

Authors Grady Hendrix and Jennifer Roberson have filed a class action lawsuit against Apple, alleging the company used pirated books to train its AI models. They claim Apple utilized the Books3 dataset, which contains copyrighted works without permission or compensation, to train its OpenELM and Foundation Language Models. The lawsuit seeks damages and an injunction, arguing Apple's AI models compete with the authors' works.

Sources

• CrowdStrike and Meta launch open source AI benchmarks for SOC
• CrowdStrike, Meta Launch New AI Security Tool
• Crowdstrike and Meta just made evaluating AI security tools easier
• The Future of AI in Healthcare: Best Practices for Responsible Innovation
• URAC Launches First-Ever Health Care AI Accreditation Program
• Soul Machines Launches Workforce Connect App to Humanize Automated Enterprise Workflows on the ServiceNow AI Platform
• Google DeepMind Detoxifies Toxic Data to Extend AI Training Resources
• Publishers and advertisers face new AI agent oversight hurdles
• Amazon bolsters AI agent push with new executive hires, according to internal memos
• Story Mining: The Essential PR Pro Skill for the AI Era
• AI Charts Course for Smarter Naval Sustainment in Canada
• Ubersuggest Puts AI Search Power in Every Marketer’s Hands with New Features
• Apple hit with class action lawsuit for using authors’ works in AI training