Before you get too far down the road of becoming an AI-driven leader, it's worth taking steps to ensure you are familiar with the potential potholes ahead of you.

This guide offers a brief summary of Dr. Immaculate Motsi-Omoijiade's comprehensive masterclass on AI risks and regulations.

With AI's help (thanks Claude!) we've synthesised key insights on the current threat landscape, regulatory frameworks, and practical implementation steps that leaders can take to ensure responsible AI adoption in their organisations.

As Mac told the masterclass, AI is akin to a "teenager" in sophistication - we simply can't treat AI like a novelty anymore.

AI risks typically fall into two categories:

Long-term concerns include extinction-level events, workforce displacement, and monopolistic control of AI capabilities.

Short/Mid-term risks focus on immediate business challenges. A 2024 Deloitte survey of leaders listed perceived risks in order:

• Security vulnerabilities (86%)
• Surveillance implications (83%)
• Privacy concerns (83%)
• Legal exposures (80%)
• Regulatory uncertainty (79%)
• Reliability issues (78%)
• Accountability gaps (75%)
• Bias and discrimination (71%)

• Bias in Facial Recognition: In one test, AWS's 'Rekognition' showed 40% of false matches involving people of colour, highlighting how embedded biases can lead to discriminatory outcomes
• Legal Vulnerabilities: Wrongful arrests based on faulty AI matches demonstrate the real-world consequences of algorithmic errors
• Dataset Limitations: Research confirms that facial recognition datasets predominantly feature lighter-skinned subjects, directly contributing to higher misclassification rates for darker-skinned individuals

Deepfake technology represents one of the most pressing AI risks today, with far-reaching implications for business security and social trust:

• Alarming Frequency: In 2024, deepfake attacks occurred every 5 minutes, creating unprecedented challenges for information verification and identity protection
• Explosive Growth: Digital document forgeries increased by 244% year-over-year, signaling a dramatic rise in sophisticated falsification capabilities
• Public Anxiety: There is widespread concern about deepfake misuse, with many individuals (including Mac herself!) struggling to differentiate between authentic and manipulated content
• Detection Challenges: Current identification tools remain inadequate, frequently generating both false positives (flagging genuine content as fake) and false negatives (missing sophisticated fakes)
• Identity Fraud Risk: Deepfakes enable increasingly convincing impersonation, creating new vectors for social engineering and targeted fraud schemes

Familiarise yourself with emerging frameworks:

• EU AI Act, provisions
• Australia's mandatory guardrails and voluntary AI safety standards.

Stay ahead by monitoring global regulatory trends and adapting proactively rather than reactively.

Don't wait for regulation to catch up – establish your own governance:

• Adopt industry-specific codes of conduct
• Pursue relevant AI certification programs
• Implement ethics labeling for AI systems

Address the "black box" problem in AI:

• Prioritise explainable models where possible
• Implement tools like LIME and SHAP to interpret complex systems (these explain AI decisions by showing each feature's importance)
• Communicate clearly about how AI decisions are made

Make fairness a technical requirement:

• Test continuously for biases during development and deployment
• Implement correction models to mitigate demographic skews
• Create diverse data training sets

Develop governance structures that define:

• Who is responsible for AI outcomes
• How concerns about discrimination are addressed
• Documentation practices for decision processes
• Regular auditing schedules

Protect AI systems from exploitation:

• Implement context-specific cybersecurity measures
• Conduct regular vulnerability assessments
• Develop response protocols for AI-specific threats

Focus on key risk areas:

• Intellectual property and copyright considerations
• Liability frameworks for autonomous systems
• Data protection and privacy requirements
• Anti-discrimination and fairness standards

Mac told the masterclass that the most effective AI governance emerges from multiple stakeholders working together:

• Engage with regulatory bodies to shape sensible frameworks
• Partner with industry peers on standard-setting
• Connect with academic institutions researching ethical AI
• Listen to public concerns about AI implementation

By taking these proactive steps, leaders can mitigate AI's most significant risks while harnessing its great potential.

Want to keep up with the latest ideas and thinking from the brightest minds in the world?

Download our blockbuster 2025 Program! Growth Faculty’s live virtual learning sessions feature some of the world's most influential thought leaders in business and leadership, many of them focusing on AI, change leadership, and adaptable leadership.

Events are added regularly, so keep up to date by signing up for our newsletter and see what’s on offer for Growth Faculty members.

Growth Faculty also offers a great value Enterprise plan for larger companies.