Content Fans

Category

Business & Ethical AI

Pieces on AI’s impact on business processes, ROI, leadership decisions, plus the risks, ethics, and reliability of these technologies.

272 articles • Page 4 of 19

Enterprises build Codex playbooks for AI governance, compliance by 2026

Enterprises build Codex playbooks for AI governance, compliance by 2026

Companies using Codex agents may struggle because there is no clear guide for making governance playbooks. Sources suggest that a playbook helps link policy and controls directly into development, which might reduce risks and speed up audits. Most organizations use a mix of NIST AI RMF 1.0 and the EU AI Act for their oversight, and experts believe a playbook should cover areas like agent inventory, risk levels, and response steps. Guidelines recommend building oversight into existing pipelines and keeping logs for audits. Playbooks may need regular updates after incidents to stay effective and follow new rules.

Anthropic Urges Human Oversight, Layered Defenses for AI-Authored Code

Anthropic Urges Human Oversight, Layered Defenses for AI-Authored Code

Anthropic warns that current safety measures for AI-generated code may not be enough, and it urges companies to use human oversight with several layers of security. Its guidance suggests humans should review and approve all important changes, while keeping logs and following clear procedures in case of problems. Anthropic also recommends starting with small pilot projects, measuring risks, and only expanding once controls seem reliable. These steps may help organizations meet new laws in the EU and US that require detailed tracking and transparency for high-risk AI systems.

Enterprises Adopt AI Governance Playbooks to Manage LLM Risks

Enterprises Adopt AI Governance Playbooks to Manage LLM Risks

Enterprises are increasingly adopting AI governance playbooks to manage risks from large language models (LLMs), as they try to balance productivity and compliance. Only about 21 percent of firms reportedly had formal generative-AI policies by mid-2025, which suggests that many organizations may still need structured guidance. Best practices appear to include combining general standards like the NIST AI Risk Management Framework with specific controls for LLMs, such as prompt-injection defenses and artifact tracking. Playbooks often recommend careful review of generated code, control gates at each workflow step, and strong artifact management. Automation and visible governance may help organizations both improve compliance and make work easier for teams.

IBM and Northflank Detail Safe AI Code Deployment Checklist

IBM and Northflank Detail Safe AI Code Deployment Checklist

IBM and Northflank share advice for safely deploying AI-generated code in companies. They suggest starting with small pilot projects, using strict testing and security checks before expanding to more teams. Human oversight and clear tracking of code changes appear to be important for meeting legal rules and catching problems early. Teams may want to wait until defect and security rates are low before wider rollout. While this approach does not guarantee perfect results, experts suggest it may help make using AI-generated software safer and more reliable.

California's new AI law mandates dataset disclosure, content-origin markers by 2026

California's new AI law mandates dataset disclosure, content-origin markers by 2026

California's new AI law, starting in 2026, will require developers with over 1 million users to share summaries of their training data and to mark where AI-generated content comes from. This may help address concerns about unclear authorship and the reuse of existing material by AI, but some risks, like data leaks and loss of original context, remain. Legal experts say that U.S. copyright still generally needs meaningful human input, so pure machine output often does not qualify. Companies are advised to use safeguards like agent registries and clear labeling of AI involvement. Experts suggest that while these rules are a first step, creative use of AI content may outpace policy, so careful tracking and governance are important.

Enterprises Face 5 Big Roadblocks Adopting Agentic AI Workflows

Enterprises Face 5 Big Roadblocks Adopting Agentic AI Workflows

Enterprises may face big challenges when trying to use agentic AI for automating workflows instead of using many single-user apps. Experts suggest that the main difficulties are not the AI models themselves, but issues with integration, control, and changing how people work. Many pilot projects stall because of problems with connecting to old systems, missing audit trails, unclear controls, and vendor lock-in. Successful adoption seems to need careful planning for both technology and organization, such as clear rules, secure integrations, and staff training. If these steps are followed, enterprises might be able to move to more efficient, automated workflows with less risk.

New Report Details How Financial Regulators Audit AI Decisions

New Report Details How Financial Regulators Audit AI Decisions

A new report suggests that public trust in AI remains low, with only about 46% of people willing to trust AI and 70% believing regulation is needed. Risk-based governance frameworks, like the NIST AI RMF and the EU AI Act, may help by requiring ongoing monitoring and human oversight for high-risk AI systems. Evidence shows that people want proof of how AI decisions are made, not just promises, so documentation and audit trails are becoming more important. In finance, regulators now focus on tracking data, decisions, and human approvals, which might become common in other high-risk areas. Experts suggest that organizations aligning with these practices and maintaining clear records may have an advantage as rules become stricter in 2026.

GameDiscoverCo Warns Studios on AI "Shovelware" and Data Leaks

GameDiscoverCo Warns Studios on AI "Shovelware" and Data Leaks

GameDiscoverCo warns that game studios may face two problems with AI: sensitive data might leak, and stores could fill up with low-quality AI-made games. Some studies suggest that AI tools have limits, and many players might avoid games made with AI. There is a risk that AI changes or mixes up important information, making it harder for real indie games to be noticed. Studios appear to be testing ways to protect their data and control what AI can use, instead of waiting for new laws. The newsletter suggests that balancing automation with human checks may help studios use AI safely without losing trust or quality.

OpenAI's Sora shutdown reveals $1 million daily GPU burn

OpenAI's Sora shutdown reveals $1 million daily GPU burn

OpenAI shut down its Sora project, which may have been using about $1 million in GPU costs per day. This move suggests that investors want companies to focus expensive computing resources on projects with clearer ways to make money. Experts now recommend that AI costs be managed throughout the model's life, from picking the right use case to watching spending after launch. Suggestions include using smaller models, optimizing training, and strict governance to link spending to business value. Companies may also save money by using a mix of different cloud computing options and closely monitoring which projects to invest in.

BCG: Agentic AI Forces Enterprise Redesign, Not Just New Models

BCG: Agentic AI Forces Enterprise Redesign, Not Just New Models

BCG's research suggests that just adding smart AI models does not usually change how a company works. Instead, companies may need to rethink how people and AI share tasks, set rules, and handle decision-making. Early reports show that agentic AI, which can plan and learn, may require changes to job roles, team structures, and investment strategies. Evidence from firms like Lenovo appears to show the real challenge is changing management and governance, not just adding new technology. The value from agentic AI might only appear if organizations redesign workflows and learn to manage these new AI agents as active team members.

AI Adoption Faces Bottleneck as Public Trust, Governance Lag

AI Adoption Faces Bottleneck as Public Trust, Governance Lag

AI progress is moving quickly, but many institutions are slow to adopt it because they are unsure if controls and governance are strong enough. Studies suggest that weak data governance, not technical ability, may be what holds back wider use, especially in areas like finance and public infrastructure. Public trust appears to be limited, and some experts warn that a lack of transparency might stop people from accepting AI. New laws and frameworks, such as the EU AI Act and ISO/IEC 42001, aim to improve oversight, but many organizations still seem to be in early stages of building trust. Evidence suggests adoption might increase most where technical advances are paired with clear records, strong controls, and public involvement.

White House briefs AI developers on 90-day model review plan

White House briefs AI developers on 90-day model review plan

The White House recently told major AI developers about a voluntary plan that may let government agencies review new AI models for up to 90 days before they are released. The main framework for these reviews, announced in March 2026, is not a binding rule but signals a shift toward federal standards and may become expected for big projects. Developers are encouraged to build internal review processes, follow security testing steps, and prepare documents for potential government checks. There are ongoing concerns about how to keep trade secrets safe during reviews, so companies may use secure methods to share only necessary information. Experts suggest that careful compliance with these reviews might help companies show they are ready for regulators and customers.

Google Urges Publishers to Adopt Edge for $4 Billion Live Ad Market

Google Urges Publishers to Adopt Edge for $4 Billion Live Ad Market

Google is urging publishers to use edge-based systems to better handle the $4 billion global live streaming ad market. Edge architecture may help serve millions of viewers at once and keep video delays very short, which could improve reliability and reduce problems like buffering. Google suggests a hybrid monetization strategy, combining different ad types for better flexibility and possibly higher revenue. Accurate forecasting of viewer numbers appears to be important for avoiding overloaded systems or missed ad opportunities. The market for connected TV ads might grow quickly in the coming years, so scaling technology may be key for publishers to succeed.

Authentic Brands Acquires Lee Denim for $1 Billion, Expands Portfolio

Authentic Brands Acquires Lee Denim for $1 Billion, Expands Portfolio

Authentic Brands Group plans to buy Lee denim for $1 billion, but the deal still needs regulatory approval and may close in late 2026. ABG says it will move Lee to a licensing model, which might help the brand expand into new product categories more quickly. Lee made about $750 million in sales in 2025, and ABG sees this as a chance to grow its denim business, which already includes Dockers and a stake in Guess. Analysts suggest this licensing plan could keep the brand visible while lowering risks, but there may be concerns if new partners do not fit well. ABG has not yet shared details about future licensees or where Lee products will be sold.

Publicis' $2.2B LiveRamp acquisition raises data control, AI concerns

Publicis' $2.2B LiveRamp acquisition raises data control, AI concerns

Publicis' planned $2.2 billion purchase of LiveRamp has started debate about who controls important marketing data and how AI will be used. Some experts suggest this deal might change the way brands and agencies share identity data, since a big piece of shared infrastructure could now belong to one company. There are concerns from competitors and regulators about possible loss of neutrality and control, and antitrust reviews may happen. LiveRamp has promised to stay open to all partners, but some worry the industry could become more divided. Advertisers may try to protect themselves by asking for more control over their data and looking for other options.