Enterprises are struggling to secure their AI deployments, with 90% of enterprise AI usage going untracked, ungoverned, or unsecured, according to FireTail's State of AI Security 2026 report. Only 29% of organizations have controls in place to secure their AI deployments, leaving them vulnerable to AI-related breaches.
IBM's 2026 report found that most AI-related breaches do not involve defeating the AI model itself, but rather exploiting exposed APIs or misconfigured cloud buckets. The average breach costs $6 million, up 35% from the previous year. Scale AI is working to address these challenges, having recently secured $6 million in funding.
Rogue Studios is using AI to generate sophisticated adult content, while Accenture emphasizes the need for companies to understand the economics of AI and manage token consumption to achieve measurable returns. The rise of AI is also causing college students to rethink their majors, with some switching to fields less likely to be automated.
Companies are also struggling with shadow AI, with 43% of security incidents last year attributed to unauthorized AI tools. Experts stress the need for robust AI governance and comprehensive asset mapping to manage exposure and ensure compliance. As AI continues to evolve, security leaders must communicate Agentic AI risk effectively and set guardrails for safe experimentation.
Heidi, an AI care partner, has built production-ready AI for healthcare at scale, using MongoDB to handle diverse medical data and connect with AI workflows. Cornell Tech has also announced new faculty members with expertise in AI, machine learning, and related fields. A Dartmouth study found that AI can help doctors sound more empathetic, but it also has limitations.
The next China shock is expected to come from open-source AI, with China promoting its own open-source AI model, PaddlePaddle, which is gaining traction globally. As the AI landscape continues to shift, companies and organizations must adapt to stay ahead.
Key Takeaways
["90% of enterprise AI usage is untracked, ungoverned, or unsecured, according to FireTail's State of AI Security 2026 report.", "The average AI-related breach costs $6 million, up 35% from the previous year, according to IBM's 2026 report.", 'Scale AI recently secured $6 million in funding to address AI security challenges.', 'Most AI-related breaches do not involve defeating the AI model itself, but rather exploiting exposed APIs or misconfigured cloud buckets.', 'Companies are struggling with shadow AI, with 43% of security incidents last year attributed to unauthorized AI tools.', 'Accenture emphasizes the need for companies to understand the economics of AI and manage token consumption to achieve measurable returns.', 'The rise of AI is causing college students to rethink their majors, with some switching to fields less likely to be automated.', 'Heidi, an AI care partner, has built production-ready AI for healthcare at scale, using MongoDB to handle diverse medical data and connect with AI workflows.', 'Cornell Tech has announced new faculty members with expertise in AI, machine learning, and related fields.', 'The next China shock is expected to come from open-source AI, with China promoting its own open-source AI model, PaddlePaddle.']AI Security: Most Enterprise AI Usage Untracked
FireTail's State of AI Security 2026 report reveals that 90% of enterprise AI usage is untracked, ungoverned, or unsecured. Only 29% of organizations have controls in place to secure their AI deployments. The report highlights a significant gap between AI adoption and security controls. AI security incidents have increased, with data exfiltration being the largest category. The report suggests that existing controls are not effective in securing AI deployments.
AI-Related Breaches Often Skip the Model
IBM's 2026 report finds that most AI-related breaches do not involve defeating the AI model itself. Instead, attackers often reach the AI system through exposed APIs or misconfigured cloud buckets. Compromised APIs and cloud misconfigurations are the most common ways in. Only 40% of organizations breached through AI had access controls on their models and data. The average breach costs $6 million, up 35% from the previous year.
Rogue Studios: AI-Generated Adult Content
Rogue Studios, a new film studio, aims to create sophisticated AI-generated adult content. The studio's founders, including a former adult film producer and a tech entrepreneur, want to create a new kind of adult entertainment that focuses on storytelling and character development. They plan to use AI to generate complex and nuanced content that is more like high-end art house films than cheap adult content.
AI Tokenomics: A New C-Suite Discipline
Accenture suggests that companies are pouring billions into AI, but many are only seeing half the picture: the cost. The real value of AI lies in its ability to deliver business value. Tokenomics, the discipline of linking AI consumption to value, is a new C-Suite mandate. Companies need to understand the economics of AI and manage token consumption to achieve measurable returns.
AI Changing College Majors
The rise of AI is causing college students to rethink their majors. Some students are switching to majors that are less likely to be automated by AI. Researchers at the St. Louis Fed found that AI is changing the job market and the skills required for certain careers. Some schools are offering new majors and programs focused on AI and related fields.
Shadow AI Risk
Companies are struggling to manage the risks associated with shadow AI, which refers to the use of unauthorized AI tools by employees. Shadow AI contributed to 43% of security incidents last year, doubling the previous figure. Experts emphasize that 70-75% of AI risk stems from usage risk, rather than inherent model flaws. Companies need to implement robust AI governance and comprehensive asset mapping to manage exposure and ensure compliance.
Agentic AI Security Awareness
The rapid emergence of agentic AI is fundamentally changing how colleagues work. Autonomous agents are executing multi-step workflows, connecting to enterprise APIs and reading internal databases. Traditional annual security awareness training is not sufficient for the agentic era. Security leaders need to communicate Agentic AI risk in plain language and set guardrails for safe experimentation.
Building Production-Ready AI for Healthcare
Heidi, an AI care partner, built production-ready AI for healthcare at global scale. The company uses MongoDB to handle diverse medical data and connect with AI workflows. Heidi's architecture is designed to be highly secure, auditable, and compliant with healthcare regulations. The company uses AI to improve patient care and streamline clinical workflows.
Cornell Tech's New Faculty
Cornell Tech has announced six new faculty members with expertise in AI, machine learning, and related fields. The new faculty members will focus on developing new AI systems that can learn, reason, and solve problems more effectively. Their research has the potential to revolutionize various industries and improve the lives of people around the world.
Dartmouth Study on AI in Healthcare
A Dartmouth study found that AI can help doctors sound more empathetic, but it also has limitations. The study used AI large language models to draft responses to patient messages, but the models introduced errors and over-diagnosed. The study suggests that AI has the potential to improve patient care, but it needs to be used carefully and with human oversight.
The Next China Shock: Open-Source AI
The next China shock will come from open-source AI, according to a report. China is promoting its own open-source AI model, PaddlePaddle, which is gaining traction globally. The adoption of PaddlePaddle could lead to a shift in the global balance of power and influence in the AI sector.
Sources
- FireTail State of AI Security 2026: Adoption Has Outpaced Control
- Most AI-Related Breaches Skip the Model, IBM's 2026 Report Finds
- This R-Rated Film Studio Wants to Be the HBO of AI
- AI Tokenomics: The New C-Suite Discipline
- AI is making college students change their majors
- Companies Cannot Price The Shadow AI Risk They Cannot See
- Building Agentic AI security awareness beyond the security team
- How Heidi built production-ready AI for healthcare at global scale
- Cornell Tech’s new faculty are changing how AI learns, reasons, solves problems
- Dartmouth study finds AI can help doctors sound more empathetic, but overdiagnoses and overtreats
- The next China shock will come from open-source AI
Comments
Please log in to post a comment.