> Markdown version of [/videos/624-the-shadows-that-follow-the-ai-generative-models](https://www.wearedevelopers.com/videos/624-the-shadows-that-follow-the-ai-generative-models). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# The shadows that follow the AI generative models

Stack Overflow restricted AI answers due to unpredictable hallucinations. Uncover the hidden operational vulnerabilities of LLMs and learn how to build robust ethical safeguards for your engineering team.

- **Speakers:** Cheuk Ho
- **Event:** World Congress 2023
- **Published:** August 11, 2023
- **Duration:** 28:30
- **URL:** https://www.wearedevelopers.com/videos/624-the-shadows-that-follow-the-ai-generative-models

## Summary

Generative AI tools like ChatGPT, Stable Diffusion, and GitHub Copilot have rapidly reshaped developer workflows, offering powerful capabilities from code assistance to automated project management. However, behind these technical leaps lie significant operational and ethical vulnerabilities. The rush to deploy large language models (LLMs) often obscures foundational flaws, including unpredictable hallucinations, prompt injection attacks, and the insidious degradation of accuracy over time. These unreliability factors have led platforms like Stack Overflow to temporarily restrict AI-generated answers, highlighting the critical need for source verification and rigorous input validation in developer environments.

Beyond technical correctness, generative AI introduces profound ethical liabilities deeply rooted in biased training data and ambiguous content ownership. Real-world applications consistently surface algorithmic bias, whether inadvertently enforcing gender stereotypes in language translation or altering racial features in professional imaging. More severe consequences arise when malicious actors exploit models for non-consensual deepfakes and automated misinformation. This lack of transparency forces engineering teams to confront difficult questions about accountability: when an AI generates harmful or illegal content, the responsibility remains dangerously blurred among the prompt user, the model creator, and the platform provider.

Addressing these "shadows" requires developers and technology leaders to actively engage in the evolving landscape of AI governance, such as navigating the EU AI Act while sustaining open-source innovation. Engineering teams are encouraged to prioritize safety by pursuing specialized roles in AI ethics, utilizing resources like the Linux Foundation's ethics courses, and championing secure open-source supply chains. By striking a balance between technological progress and actionable regulatory constraints, the software industry can build robust safeguards against malicious behavior and ensure that emerging AI applications remain trustworthy.

**Keywords:** generative AI models, LLM hallucination risks, AI prompt injection vulnerabilities, deepfake content regulations, algorithmic bias mitigation, AI output liability, open-source supply chain security, software engineering AI ethics, EU AI act compliance, model accuracy degradation, developer ecosystem AI bans, coding assistant limitations, training data bias, AI governance frameworks, open-source AI regulation

## Chapters

1. **Introduction to generative AI and content warnings** (01:35) — An overview of generative AI models alongside content warnings regarding sensitive topics.
1. **Basics of generative AI and prompt interactions** (03:50) — How generative AI models process training data and user prompts to produce new content.
1. **Examples of generative AI applications in media and code** (05:34) — A look at AI-generated images, music, code, and project management tools.
1. **Evaluating the accuracy and correctness of generated content** (09:22) — The challenges of relying on AI for accurate technical answers and the risks of model hallucination.
1. **Security risks involving prompt injection attacks** (14:26) — How prompt injections allow users to manipulate model outputs similar to SQL injection attacks.
1. **Determining ownership and responsibility for generated content** (15:54) — The ethical dilemmas around assigning responsibility for copyright and inappropriate outputs in generated content.
1. **Addressing inherent biases in large language models** (18:01) — Examples of how training data perpetuates racial and gender biases in generated outputs.
1. **Harmful applications including deepfakes and non-consensual imagery** (21:38) — The illegal use of generative AI to create non-consensual imagery and fraudulent identities.
1. **Emerging regulations and cultivating careers in AI ethics** (23:47) — Current regulatory efforts and opportunities for professionals to engage with AI ethics initiatives.
1. **Audience questions on deepfake detection, regulations, and ethics** (25:37) — A discussion on detecting deepfakes, balancing regulations with startup innovation, and mitigating systemic biases.

## Related Moments

- [Mitigating the inherent challenges of generative AI tools](https://www.wearedevelopers.com/videos/844-enter-the-brave-new-world-of-genai-with-vector-search) (from "Enter the Brave New World of GenAI with Vector Search")
- [Identifying and hardening against generative AI risks](https://www.wearedevelopers.com/videos/1544-responsible-ai-microsoft-governance-standards-learnings) (from "Responsible AI @ Microsoft - Governance, Standards, Learnings")
- [Addressing psychological safety and ethical risks of AI adoption](https://www.wearedevelopers.com/videos/1950-the-scrum-master-as-an-orchestrator-guiding-human-ai-collaboration-in-modern-teams) (from "The Scrum Master as an Orchestrator: Guiding Human–AI Collaboration in Modern Teams")
- [Core challenges facing the generative AI developer ecosystem today](https://www.wearedevelopers.com/videos/1116-the-data-phoenix-the-future-of-the-internet-and-the-open-web) (from "The Data Phoenix: The future of the Internet and the Open Web")
- [Moving from secure software behavior to trustworthy AI](https://www.wearedevelopers.com/videos/1948-building-trustworthy-ai-in-industry-beyond-traditional-cybersecurity) (from "Building Trustworthy AI in Industry: Beyond Traditional Cybersecurity")
- [Balancing AI regulation with technological innovation in human resources](https://www.wearedevelopers.com/videos/1356-from-learning-to-leading-why-hr-needs-a-chatgpt-license) (from "From Learning to Leading: Why HR Needs a ChatGPT License")

## Related Articles

- [WWC24 Talk - Scott Hanselman - AI: Superhero or Supervillain?](https://www.wearedevelopers.com/magazine/469-wwc24-talk-scott-hanselman-ai-superhero-or-supervillain)
- [Should AI be Regulated? The Arguments For and Against](https://www.wearedevelopers.com/magazine/271-should-ai-be-regulated-the-arguments-for-and-against)
- [Panel Discussion: Responsible AI in Practice - Real-World Examples and Challenges](https://www.wearedevelopers.com/magazine/488-panel-discussion-responsible-ai-in-practice-real-world-examples-and-challenges)
- [How to Use Generative AI to Accelerate Learning to Code](https://www.wearedevelopers.com/magazine/530-how-to-use-generative-ai-to-accelerate-learning-to-code)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Staff Developer Advocate, GitHub Security Lab](https://www.wearedevelopers.com/jobs/ext/1921051-staff-developer-advocate-github-security-lab) at **GitHub**
- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [AI Operations Manager (all genders)](https://www.wearedevelopers.com/jobs/48263-ai-operations-manager-all-genders) at **envelio**
- [AI & Machine Learning Engineer (all genders)](https://www.wearedevelopers.com/jobs/48217-ai-machine-learning-engineer-all-genders) at **msg**