Top 10: AI Regulations and Compliance Issues
As AI develops across the world, it’s becoming harder and harder to employ effective regulations.
Gen AI systems for example, are transforming industries from healthcare to finance, as governments worldwide grapple with balancing innovation against emerging risks.
The EU leads with its AI Act, imposing strict obligations on high-risk AI applications, while the US pursues a fragmented approach combining federal executive orders with state-level legislation.
Meanwhile, the UK adopts principles-based regulation, seeking to maintain its position as an AI hub.
This regulatory divergence creates particular challenges for multinational technology companies deploying AI systems across borders – and the cost of compliance has become a strategic consideration, with organisations investing millions in governance frameworks, legal expertise and technical solutions.
10. Cross-border data transfers
Why it’s an issue: Global data flows for AI conflict with national data sovereignty rules
Company tackling this: Duality Technologies
AI development requires vast datasets that often span multiple jurisdictions, creating conflicts with data sovereignty regulations including GDPR.
For example, the invalidation of the EU-US Privacy Shield in 2020 demonstrated the complexity of compliant transfers.
This means that organisations must navigate Standard Contractual Clauses and Binding Corporate Rules, requiring detailed Transfer Impact Assessments.
Duality Technologies is one company that addresses these challenges through its SecurePlus platform, which uses homomorphic encryption to enable data collaboration without exposing sensitive information.
9. AI Governance and risk management
Why it’s an issue: Proactive frameworks are crucial for managing evolving AI risks and compliance
Company tackling this: KPMG
AI adoption introduces cyber threats, operational failures and regulatory non-compliance risks.
As a result, organisations face financial penalties without adequate governance frameworks.
Now, managing AI risks has evolved from ad-hoc activities to structured methodologies requiring dedicated tools.
Tackling this challenge, KPMG offers a trusted AI framework encompassing reliability, security, safety, privacy, explainability, fairness and accountability.
The framework aligns with the NIST AI Risk Management Framework and EU AI Act.
8. Fragmented global regulatory landscape
Why it’s an issue: Inconsistent laws create complex, costly and uncertain compliance for global firms
Company tackling this: TrustArc
TrustArc provides AI governance solutions through its PrivacyCentral platform, enabling adherence to various regulations including the EU AI Act and Colorado AI Act.
The platform incorporates standards like the NIST AI Risk Management Framework and OECD AI Principles.
Platforms like this must exist as global AI regulatory environments vary significantly across jurisdictions.
For instance, the EU enacted an AI Act whilst the US relies on state-level laws and federal executive orders.
Meanwhile the UK pursues a principles-based framework.
Yet this fragmentation creates complex compliance requirements for multinational organisations.
7. Deepfakes and misinformation
Why it’s an issue: AI-generated fakes undermine trust and enable fraud/manipulation
Company tackling this: Intel
AI-generated deepfakes threaten public trust, democratic processes and corporate security.
Synthetic media including manipulated videos, images and audio are increasingly sophisticated and difficult to distinguish from authentic content.
Governments respond with legislation, such as New Hampshire criminalising malicious deepfakes and California’s Defending Democracy from Deepfake Deception Act.
In response, Intel developed FakeCatcher, the first real-time deepfake detector analysing biological signals to determine video authenticity.
Running on Intel processors, the technology supports multiple real-time detection streams.
6. Intellectual property rights
Why it’s an issue: AI training and output challenge existing copyright and ownership laws
Company tackling this: Adobe
Gen AI creates complexities for intellectual property rights, particularly copyright ownership and infringement – yet most copyright and patent laws do not address AI’s role in authorship or inventorship.
Debates centre on whether using copyrighted materials to train AI models constitutes infringement and who owns copyright for AI-generated content.
High-profile lawsuits include movie studios suing Midjourney for allegedly copying protected characters.
Tackling this copyright challenge, Adobe invests in companies like Truepic, supporting authentication technologies combating deepfakes.
Adobe also participates in the Coalition for Content Provenance and Authenticity standard, providing verifiable details about digital content origin and edits.
5. AI safety and security
Why it’s an issue: AI introduces novel cyber threats and operational risks
Company tackling this: Palo Alto Networks
AI deployment creates cybersecurity threats including prompt injections, model inversion and data poisoning that compromise AI accuracy and integrity.
As companies are grappling to protect themselves, companies such as Palo Alto Networks provide products and services safeguarding AI systems against novel cyber threats, ensuring model integrity and infrastructure resilience.
US authorities including the FBI, NSA and CISA issued joint guidance warning against security threats impacting AI training data and outputs to tackle the problem.
4. Accountability and human oversight
Why it’s an issue: Autonomous AI lacks clear human responsibility and control
Company tackling this: Open AI
AI systems gaining autonomy create challenges establishing responsibility and ensuring human oversight.
As a result, UNESCO principles mandate that AI systems must not usurp human responsibility and accountability.
Human-in-the-Loop approaches embed human judgement in AI operational workflows, particularly for high-stakes domains where errors have consequences.
Similarly, Open AI emphasises accountability and human oversight through open-sourced safety research and collaboration with governments on AI governance frameworks.
The company focuses on building steerable, interpretable AI systems aligned with human values, ensuring human responsibility remains paramount.
3. Transparency and explainability
Why it’s an issue: Opaque AI decisions erode trust and hinder accountability and auditability
Company tackling this: Google
The black box nature of AI systems obscures decision reasoning, eroding user trust and impeding accountability – leading UNESCO global standards and emerging regulations like the UK AI Bill emphasise transparency and explainability importance.
Explainable AI enables compliance by facilitating bias detection and correction whilst ensuring AI judgements in healthcare and finance are comprehensible.
Google’s DeepMind division is one enterprise leading ethical AI development emphasising explainable AI in projects.
Google’s Gemini models also have reasoning capabilities contributing to understandable AI outputs.
2. Data privacy and protection
Why it’s an issue: AI processes vast data, risking breaches and misuse of personal info
Company tackling this: Microsoft
Microsoft addresses data privacy and protection challenges in many ways including through its Purview suite delivering unified data governance capabilities.
The platform offers automated data discovery and classification with integrated security solutions including Data Loss Prevention and Information Protection across on-premises, multi-cloud and SaaS environments.
Microsoft’s Responsible AI principles also embed privacy and security by design throughout AI system lifecycles.
Enterprises are innovating in this way as AI systems process immense data volumes raising privacy and protection concerns regarding sensitive personal information.
As a result, GDPR compliance is paramount, with EDPB guidelines on data transfers and training courses for Data Protection Officers.
Data breaches and personal data misuse rank among top business concerns.
1. Algorithmic bias and fairness
Why it’s an issue: Unfair AI leads to discrimination, legal risks and reputational damage
Company tackling this: IBM
Algorithmic bias from unrepresentative training data results in discriminatory outcomes in hiring processes and loan approvals.
Such biases carry legal and reputational risks making fairness pursuit a business imperative. Regulatory bodies enforcing the EU AI Act and US Algorithmic Accountability Act mandate bias mitigation strategies.
A company leading in solutions to algorithmic bias and fairness problems is IBM – that provides the open-source AI Fairness 360 toolkit – offering 70 fairness metrics and 10 bias mitigation algorithms detecting and reducing discrimination throughout AI application lifecycles.
Furthermore, IBM’s AI Ethics Board – co-chaired by Chief Privacy Officer Christina Montgomery and AI Ethics Global Leader Francesca Rossi, reviews AI use cases ensuring alignment with fairness principles.
Explore the latest edition of AI Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.
Discover all our upcoming events and secure your tickets today.
Also sign up to our free weekly newsletter for the latest insights and stories straight into your inbox.
AI Magazine is a BizClik brand
