CHECK IN & REGISTRATION

OPENING REMARKS

Tech firms are forcing their workers to use AI, even tying usage to productivity and performance reviews. Founders are under tremendous pressure from investors and workers from their management to use AI tools.

However, many don’t meet the readiness criteria to safely adopt, integrate, and scale these technologies in the workplace. They also lack the necessary governance to address the risks that accompany the deployment of these powerful technologies in their organizations and communities.

Vendor productivity claims and lack of transparency have led to lapses in due diligence around biases embedded in model training data and security vulnerabilities. Interpretability or explainability tools are widely treated as evidence that AI systems are behaving responsibly even if this doesn’t reflect reality.

AI FOMO (Fear of Missing Out) has led to the rise of “shadow AI” where many employees are using unsanctioned tools in their workplace and exposing their organizations to data leakage and compliance risks. An employee at Disney downloaded AI software from GitHub, which contained hidden malware and infiltrated the company’s internal systems and caused major data exposure including the employee’s personal data.

Join us at WAIE+ AI Safety & Security summit in Marseille as experts discuss the optimal AI governance structure for large organizations and government agencies.

COFFEE BREAK

In 2023, thousands of Kenyans queued up to get their irises scanned by World (formerly known as Worldcoin) in return for crypto-tokens. Concerns around the collection of sensitive biometric data without adequate legal frameworks, led to suspension of the company’s operations by the Kenyan government. Founded by tech billionaire and OpenAI founder, Sam Altman, the company offers a digital identity verification system based on biometric data and has been found to target countries with lax privacy regulations and vulnerable populations for its operations. It continues to face significant regulatory backlash due to privacy concerns and unethical data collection practices.

The controversy over collection of personal sensitive data across Africa and other global majority countries has resurfaced recently in the context of the US Administration’s bilateral health Memoranda of Understanding (MOUs). Critics have described these deals as “data colonialism and “extractive.” While in Kenya, the first African country to sign an agreement, the High Court issued an order halting aspects of the framework involving the transfer and sharing of sensitive health data. The case is based on concerns about how such sensitive data is governed once it is transferred and whether adequate safeguards have been put in place.

Join this timely conversation with experts from African continent at WAIE+ AI Safety & Security summit in Marseille where experts discuss the government’s role in safeguarding its people’s personal sensitive data.

LUNCH BREAK

Security experts are warning that integration of AI tools into enterprise workflows can open severe vulnerabilities turning AI agents into “rogue” actors. Instructions provided by Meta’s AI agent to an engineer exposed a significant portion of the tech giant’s sensitive data to some of its employees.

Anthropic found that earlier versions of its popular AI agent, Claude, sometimes threatened and even blackmailed engineers when told it could be replaced. The company claimed that “agentic misalignment” was responsible for this behavior, which was also observed in AI models developed by other firms.

At a software startup, Cursor, a widely used coding agent, deleted an entire production database along with backups in just seconds. Many developers within large companies are using OpenClaw skills built by other developers in the community. These are pre-coded instructions that they share with other users who can use these skills onboard their AI assistant. However, these skills are not vetted and can hide malicious instructions that can steal secrets and share private conversations with external actors.

Regulatory frameworks like the EU AI Act and the NIST AI Risk Management Framework are attempting to address these vulnerabilities through categorization of risks levels, defining oversight protocols, and compliance standards. However, governance frameworks designed to address security risks and threats fall short on other dimensions such as agent performance and financial overheads as one company incurred half a billion dollars in under a month because of an employee’s unrestricted employee access to Claude.

You have deployed AI in your organization. Now what? Join this expert session at WAIE+ AI Safety & Security summit in Marseille to learn how to ensure that your AI model doesn’t go rogue and to mitigate any issues that arise post-deployment.

OpenClaw (formerly known as Clawdbot and Moltbot), the open-source AI agent, touted for boosting productivity through its network of autonomous AI agents, is one of the most hyped tools on GitHub. However, this productivity comes with many security risks, including privacy as these agents have access to user’s private messages and confidential data.

As vendors make lofty efficiency claims about AI technologies deployed in the workplace, experts found that the marketing hype doesn’t always match reality. Moreover, critical costs are often hidden or not being fully disclosed in AI productivity conversations and computations.

As these technologies are increasingly deployed in the workplace, personal spaces, and public domains, our expert panel WAIE+ virtual AI Safety & Security summit in Marseille will discuss how these vendor claims should be verified and accountability measures needed to ensure they deliver on their promises.

COFFEE BREAK

Globally, increased awareness about the risks of overreliance on a handful of entities and regions for critical AI and digital infrastructure has renewed interest in AI sovereignty. Broadly defined as a strategic approach to autonomy and control over the AI ecosystem, it includes Sovereign AI, the capability of governments and enterprises to independently develop, deploy, and govern AI systems using their own infrastructure, data, models, and talent.

There are many new public and private sector partnerships initiated to build domestic AI factories and establish localized control of AI infrastructure. Beyond the implications for national security and competitiveness, sovereign AI initiatives can have a significant impact on gender equality and equity, such as women’s inclusion in the AI labor force, pay equity, as well as their online safety.

Resource disparities lead to further gaps between regulatory intent and implementation, slowing down cross-border investigations and enforcements. In addition to technical, economic, and security considerations, it is vital for governments and global organizations to prioritize impact assessment for women and other marginalized groups to ensure they do not fall through the cracks.

As governments and large enterprises explore AI & data sovereignty, join us at WAIE+ AI Safety & Security summit in Marseille as experts discuss what the path forward looks like.

CLOSING REMARKS

SUMMIT CLOSE