AI Rules Take Shape as Adoption Surges

Governments move to put guardrails on AI

Policymakers are racing to set rules for artificial intelligence as the technology spreads across business and daily life. The European Union has approved the AI Act, which it calls the world’s first comprehensive AI law. The United States has issued an executive order on AI safety. The United Kingdom hosted a global summit to coordinate risk standards. These efforts show a common goal: keep benefits high and harms low as AI scales.

The EU law uses a risk-based approach. It bans some applications and sets tougher duties for systems seen as high risk, such as those used in hiring, education, and critical infrastructure. It also adds obligations for general-purpose AI, including large models that power chatbots and coding tools. Fines for violations can be steep. Under the act, penalties can reach up to 7% of a company’s global annual turnover, depending on the breach. Many provisions will phase in over the next two to three years.

In Washington, the White House issued an executive order in October 2023 on “Safe, Secure, and Trustworthy AI.” It directs agencies to set testing standards, assess national security risks, and protect consumers and workers. It also tasks the National Institute of Standards and Technology (NIST) with developing new evaluation methods. The order seeks reporting from developers of the most powerful models about safety test results.

The UK convened leaders at the AI Safety Summit in November 2023. The meeting produced the Bletchley Declaration, signed by dozens of nations. It recognizes that frontier AI could pose serious risks if not managed. The UK then stood up a national AI Safety Institute to test advanced systems.

Industry speeds ahead with new tools

While rules take shape, companies are deploying AI at scale. Generative AI, which can produce text, images, and code, has moved from pilot to production in many firms. It is embedded in office software, customer support, and developer tools. Chipmakers, cloud providers, and startups are all jockeying for position.

Nvidia, the leader in AI chips, saw demand surge as data centers trained and ran large models. “A new computing era has begun,” Nvidia chief executive Jensen Huang said in 2023 as the company reported rising sales tied to AI workloads. Cloud providers have rolled out specialized hardware and services to host and optimize these models.

Some AI creators also stress caution. “I think people should be happy that we’re a little bit scared of this,” OpenAI chief executive Sam Altman told ABC News in March 2023, referring to the power of advanced models. His remark underscored a growing industry view: strong safety practices help sustain public trust.

What the rules mean for businesses

For many companies, compliance planning begins now. The EU AI Act will reach beyond Europe because many firms sell into the bloc. The law defines roles across the supply chain. It sets duties for providers, deployers, importers, and distributors.

  • Inventory and risk mapping. Companies should list AI systems they build or use and classify them by risk. That includes general-purpose systems and tools adapted for specific tasks.
  • Technical documentation. The EU law expects clear, updated documentation for high-risk systems. Firms should explain how models were designed, trained, tested, and monitored.
  • Data governance. Training data quality and bias controls are central. Steps include documenting sources, cleaning data, and measuring disparate impacts.
  • Human oversight. The law calls for safeguards, such as review workflows and clear fallback procedures when systems err.
  • Transparency to users. Systems that interact with people must disclose that they are AI. Deepfakes and synthetic media need labels in many cases.

Financial services, healthcare, and public sector users face extra scrutiny. Vendors to those sectors will need assurance programs, incident reporting, and post-market monitoring. Small firms may qualify for support and regulatory sandboxes, but they still must meet core obligations.

Workers and consumers look for protections

Labor groups and civil society organizations want clear guardrails on surveillance and automated decision-making. Unions have pressed for limits on AI use in performance tracking and hiring. In 2023, the Writers Guild of America secured contract terms that set boundaries on the use of generative AI in scripted content. Privacy advocates call for strong safeguards around biometric systems and face recognition.

The EU AI Act addresses some of these concerns. It bans practices seen as unacceptable, including certain forms of social scoring and manipulative techniques that can harm vulnerable groups. It restricts real-time remote biometric identification in public spaces, subject to narrow law-enforcement exceptions and oversight.

Consumer groups also urge transparency and a right to recourse. That includes easy ways to contest automated decisions and to reach a human when systems fail. National data protection authorities will remain key enforcers in many cases, working alongside newly designated AI regulators.

Investment and adoption trends

Despite tighter scrutiny, money is still flowing into AI tools and infrastructure. The Stanford AI Index 2024 reported that private AI investment reached roughly $67 billion in 2023 worldwide. Spending focused on model development, chips, and applied tools in sectors such as software, marketing, and biotech. Public funding for research also grew.

Early studies of workplace use show promise and limits. AI assistants can help with drafting, coding, and customer support. Some experiments report faster task completion and improved quality for routine work. But performance can drop on complex, novel tasks without careful oversight. Experts warn that overreliance can spread errors and bias. Training and clear guidelines are essential.

Key questions ahead

As rules roll out, several issues will shape the next phase:

  • Model evaluations. Governments and labs are building shared tests for safety, security, and societal impact. Agreement on methods will be critical.
  • Open vs. closed models. Policymakers are weighing how to support open science while managing risks from widely available code and weights.
  • Cross-border enforcement. Coordination among EU, U.S., UK, and others will matter. Divergent rules could raise costs and fragment markets.
  • Liability. Courts will test how existing product liability and consumer protection laws apply to AI failures and misinformation.
  • Workforce shifts. Companies will need to reskill workers and update job designs. New roles in AI safety, evaluation, and governance are emerging.

The bottom line

The message from capitals is clear: AI must be safe, fair, and accountable. The message from industry is also clear: AI is becoming a core part of computing and business operations. Those two forces are now meeting in law, standards, and contracts. Companies that invest in governance and transparency will be better placed to use AI at scale.

The timeline is tight. The EU AI Act will phase in over the next few years. U.S. agencies are writing guidance now. The UK and other countries are building testing capacity. Global standards bodies, including ISO and IEEE, are updating best practices.

The outcome will affect how people work, learn, and access services. It will also influence who wins the next wave of digital growth. As one industry leader put it, a new era has begun. The challenge for policymakers, developers, and the public is to make sure it is an era that earns trust.

MOST POPULAR