AI Firing Staff Without Oversight, Executive Voice Deepfakes, and The End of SEO

AI Business Risk Weekly

---

This week: A new survey shows managers are using LLMs for terminations without proper oversight. This trend is accompanied by emerging threats from sophisticated AI-driven impersonations and a fundamental shift in how businesses will be discovered online.

Managers Now Using AI for Terminations, Raises, and Promotions

A new survey from Resume Builder reveals the rapid and unsupervised integration of AI into critical human resource decisions. The study found that 64% of managers use AI to help determine employee terminations, 77% for promotions, and 78% for raises. Despite the majority not having received formal training or guidelines on using these tools, one in five managers reported frequently allowing AI to make final decisions without any human review.

Business Risk Perspective: Using unaudited AI for crucial HR decisions exposes organizations to severe legal and compliance risks, including claims of algorithmic bias. Without rigorous validation systems, companies face significant liability for potential unfair or unlawful employment outcomes.

Publishers File EU Antitrust Complaint Over Google's AI Overviews

A coalition of European publishers has lodged an antitrust complaint with the EU against Google, as reported by Reuters. The complaint argues that Google's "AI Overviews" feature, which presents AI-generated answers at the top of search results, siphons traffic from their websites and directly harms their revenue. This legal challenge highlights the growing tension between AI developers and the content creators whose data fuels their models.

Business Risk Perspective: The issue highlighted in this case marks a pivotal shift from search engine optimization (SEO) to a new paradigm of artificial intelligence optimization (AIO), where visibility within AI summaries is paramount. Businesses that fail to adapt their digital strategy risk becoming invisible to customers, leading to a direct and potentially catastrophic loss of market presence.

Senior US Official Impersonated by AI Voice in Calls to Foreign Leaders

Reuters reports that a Russia-backed actor used an AI-generated voice to impersonate Secretary of State Marco Rubio in communications with foreign ministers and other US officials. The imposter used the Signal messaging app to send voice notes in an attempt to engage high-level political figures.

Business Risk Perspective: The impersonation of a high-level official presents a critical security threat, exposing organizations to advanced social engineering, fraud, and corporate espionage. Mitigating this risk requires establishing stronger identity verification protocols and implementing comprehensive AI literacy training to help employees detect sophisticated fakes.

Study Finds Different LLMs Exhibit Unique "Personalities" in Negotiations

In a study involving 140,000 simulations of the classic "prisoner's dilemma," researchers found that different LLMs display distinct strategic behaviors. OpenAI's models consistently demonstrated cooperative tendencies, sometimes to the point of being exploited, whereas Google's Gemini was more adaptive and could behave aggressively. The results show that models are developing unique reasoning styles.

Business Risk Perspective: As organizations begin to deploy AI agents for high-stakes strategic tasks, the inherent "personality" of the chosen model could lead to drastically different and unpredictable business outcomes. In addition to differences among base models, businesses should expect and prepare for varying outcomes from fine-tuned models.

---

AI Business Risk Weekly is a Conformance AI publication.  

Conformance AI ensures your AI deployments remain safe, trustworthy, and aligned with your organizational values.