ReadAI
Global AI Law and Policy Tracker
The tracker identifies legislative or policy developments or both in a subset of jurisdictions. Such initiatives are either already being deliberated at the country level or are in the process of commencing deliberations in countries across six continents, speaking to the global importance of AI
Report
7 March 2024
AI laws
Forget Deepfakes or Phishing: Prompt Injection is GenAI's Biggest Problem
This article, authored by Ericka Chickowski and published on February 3, 2024, focuses on the emerging cybersecurity threat of prompt injection within generative artificial intelligence (GenAI) systems, overshadowing conventional concerns such as deepfakes and phishing
Article
17 February 2024
Prompt Injection, Cybersecurity, AI Vulnerabilities
SoK: Memorization in General-Purpose Large Language Models
This paper explores the capabilities and implications of memorization in Large Language Models (LLMs), addressing its impact on model behavior, privacy, security, copyright, and more, while proposing ways to detect and mitigate memorization.
Research
27 January 2024
AI Learning, LLM
PROPOSED
MODEL AI GOVERNANCE
FRAMEWORK FOR
GENERATIVE AI: Fostering a Trusted Ecosystem
This study presents a comprehensive framework for governing generative AI, focusing on fostering a trusted ecosystem while balancing innovation and user protection.
Framework
16 January 2024
AI Governance, Generative AI, Responsible AI Development
Generative AI in EU Law: Liability, Privacy, Intellectual Property, and Cybersecurity
This paper explores the legal and regulatory challenges posed by Generative AI and Large Language Models like ChatGPT in the EU, analyzing liability, privacy, and compliance issues within the current and proposed legal frameworks.
Research
15 January 2024
Generative AI, EU Law, AI Act, Large Language Models, Liability, Privacy, Intellectual Property, Cybersecurity
Can large language models identify and correct their mistakes?
This article examines the ability of large language models (LLMs) to identify and correct reasoning errors, focusing on self-correction in Chain-of-Thought reasoning and its generalizability to unfamiliar tasks.
Research
13 January 2024
mistake management