Generative AI
A category of AI that can create new content — text, images, code, audio, and video — by learning patterns from training data.
What Is Generative AI?
Generative AI (GenAI) refers to artificial intelligence systems capable of creating new content — text, images, audio, video, code, or synthetic data — rather than simply classifying or predicting from existing data. Generative AI models learn the statistical patterns in their training data and use those patterns to produce new outputs that are plausible and contextually relevant.
Large language models (LLMs) like GPT-4 and Claude are the most prominent form of generative AI. Image generators like DALL-E, Midjourney, and Stable Diffusion are another major category.
How Generative AI Differs from Traditional AI
Traditional AI (discriminative models): Classifies or predicts. "Is this email spam or not spam?" "What is the probability this transaction is fraudulent?" Analyses existing data.
Generative AI: Creates new data. "Write me a marketing email." "Generate an image of a futuristic city." "Summarise this contract." Produces novel outputs.
Key Generative AI Technologies
- LLMs: Generate text and code; power chatbots, copilots, and document tools
- Diffusion models: Generate images (Stable Diffusion, DALL-E, Midjourney)
- Voice synthesis: Generate realistic human speech from text
- Video generation: Create synthetic video from text prompts (Sora, Runway)
- Code generation: GitHub Copilot, Cursor, Claude — write and complete code
Generative AI Security Risks
Deepfakes: GenAI creates synthetic media that can impersonate real people — used in social engineering and fraud.
Phishing at scale: LLMs dramatically lower the cost of crafting personalised, convincing phishing emails.
Malware generation: LLMs can assist in writing malicious code, lowering the skill barrier for attackers.
Data leakage via shadow AI: Employees sharing sensitive data with consumer GenAI services.
Hallucinated outputs: AI generating plausible but incorrect or fabricated information used without verification.
The Business Opportunity — and Obligation
GenAI tools offer genuine productivity gains across writing, coding, analysis, and customer service. Organisations that govern AI well — with clear policies, approved tools, and security controls — can capture that value while managing the risks. Those that don't govern it invite shadow AI and compliance exposure.