Skip to main content
Dec 20, 2023

The week in GRC: AI can’t complete SEC filings correctly, researchers find and Icahn plans to remove Illumina directors

This week’s governance, compliance and risk-management stories from around the web

The Guardian reported that Johnny Taylor, president and CEO of the Society of Human Resource Management, said diversity, equity and inclusion (DE&I) policies within US companies will ‘come under full-out attack in 2024. It’s going to become a hot-button issue [next] year.’ The national shift to be more inclusive that followed the murder of George Floyd in 2020 and the Black Lives Matter protests that followed is fading, he said: ‘We’re already seeing companies go away from it.’ He suggested the reset could be as strong as the backlash against the ESG movement.

The reversal comes after the US Supreme Court struck down affirmative action in higher education. Although the ruling did not explicitly mention company DE&I policies, the conservative movement that brought the issue to the court has its eye on DE&I in the workplace.

– Norges Bank Investment Management (NBIM) in 2021 started to pre-disclose its voting intentions five days before any shareholder meeting, arguing that ‘a lack of information makes the market for voting advice not fully efficient’ and the Financial Times (paywall) this week reported on provisional results of the policy.

Rüdiger Fahlenbrach of the European Corporate Governance Institute, Nicolas Rudolf of the University of Lausanne and Alexis Wegerich of NBIM found that by disclosing how it plans to vote, the $1.3 tn Norwegian sovereign wealth fund could in practice very roughly treble its impact, on average. ‘Voting pre-disclosures are more effective for proposals with a higher information demand and if the large shareholder pre-discloses a decision that is not directly observable from its proxy-voting guidelines,’ they wrote.

When NBIM supports a shareholder proposal, it increases the favorable vote by 3.6 percentage points. The apparent leverage gained through transparency means NBIM is a genuine player in the proxy season.

– According to Reuters (paywall), hotel operator Wyndham Hotels & Resorts asked its shareholders to reject Choice Hotels’ takeover offer, citing regulatory review of up to 24 months and lower valuation. Choice has launched a hostile bid for Wyndham after the company repeatedly rejected previous overtures. ‘We are confident Wyndham can deliver long-term shareholder value well in excess of the $85 per share offered by Choice by continuing to execute on our existing business plan,’ said Stephen Holmes, chair of the Wyndham board.

Wyndham added that about 80 percent of its franchisee respondents said a tie-up would hurt their business and about 60 percent said they would terminate their contract in the event of a merger if they had the option, according to a survey.

Choice has said it was identifying director candidates for nomination to Wyndham’s board at its 2024 AGM and would file documents with the Federal Trade Commission to kick off a regulatory review process.

CNBC reported that the EU has opened infringement proceedings into X, previously known as Twitter, over suspected failure to combat content disinformation and manipulation. It is the first such probe under the Digital Services Act (DSA). Regulator Thierry Breton said the move is in response to suspected breaches of X’s transparency obligations and its duties to counter illegal content and disinformation. It is also in response to what the EU calls a ‘deceptive’ design of user interface, relating to its so-called blue checks.

X said in a pre-prepared statement distributed on social media that it ‘remains committed to complying with the [DSA] and is co-operating with the regulatory process. It is important that this process remains free of political influence and follows the law. X is focused on creating a safe and inclusive environment for all users on our platform, while protecting freedom of expression, and we will continue to work tirelessly toward this goal.’

The European Commission said it launched the proceedings under the DSA ‘on the basis of the preliminary investigation conducted so far, including on the basis of an analysis of the risk-assessment report submitted by X in September, X’s transparency report published on November 3 and X’s replies to a formal request for information, which, among others, concerned the dissemination of illegal content in the context of Hamas’ terrorist attacks against Israel.’

– Southwest Airlines agreed to pay a record $140 mn civil penalty over the December 2022 holiday disruption that led to 16,900 flight cancellations and 2 mn stranded passengers, according to Reuters. The US Department of Transportation settlement includes a $35 mn cash fine and a three-year mandate that Southwest provide $90 mn in travel vouchers of $75 or more to passengers delayed at least three hours getting to final destinations because of an airline-caused issue or cancellation.

Southwest has over the last year made significant technology and consumer service upgrades and other investments, including more de-icing equipment, new staffing and software and using artificial intelligence (AI) to predict network health. Southwest CEO Bob Jordan said the airline was pleased to resolve the investigation even though it did not admit wrongdoing.

‘It was a historic storm that led to a historic week of operational disruption,’ Jordan said, adding that a disruption of that magnitude ‘is not going to happen again’.

CNBC reported that, according to its research, employees who use AI at work say they are more likely to view it as a positive, with 72 percent reporting that it has made them more productive. Gen Zs (37 percent) and millennials (35 percent) are the most likely to have used AI in their jobs. Just 25 percent of Gen Xs and 17 percent of baby boomers report using AI tools such as ChatGPT at work.

Despite workers’ positive feedback about AI, however, 42 percent of employees said they’re concerned about the technology’s impact on their jobs. The outlook on AI as a job threat also varied by salary, with workers making under $50,000 more concerned (47 percent) about the technology’s impact than those making between $50,000 and $99,000 (39 percent) or $100,000 or more annually (36 percent).

CNBC also reported researchers from a start-up called Patronus AI finding that large language models (LLMs), similar to the one used by ChatGPT, often fail to answer questions derived from SEC filings. Even the best-performing AI model configuration they tested, OpenAI’s GPT-4-Turbo, which can read nearly an entire filing alongside the question, only got 79 percent of answers correct on Patronus AI’s new test.

LLMs would often refuse to answer or would ‘hallucinate’ figures and facts that weren’t in the SEC filings. The findings highlight some of the challenges facing AI models as big companies, particularly in regulated industries such as finance, look to incorporate cutting-edge technology into their operations in areas including customer service or research.

An OpenAI representative noted that the company’s usage guidelines prohibit offering tailored financial advice using an OpenAI model without a qualified person reviewing the information and require anyone using an OpenAI model in the financial industry to provide a disclaimer informing that AI is being used and detailing its limitations. OpenAI’s usage policies also say that OpenAI’s models are not fine-tuned to provide financial advice.

– Activist investor Carl Icahn said he plans to remove directors at Illumina, setting the stage for a second board challenge at the company months after shareholders elected one of his director candidates, according to Reuters. Icahn disclosed his plans without providing details in a letter to other shareholders less than 24 hours after Illumina said it would divest blood-test maker Grail.

Icahn cheered the company’s decision to divest Grail but said the job at Illumina has not been fully completed. ‘Our third goal is to remove these legacy-conflicted directors,’ he wrote in a letter that was made public in a regulatory filing.

Illumina declined to comment.

The Wall Street Journal (paywall) reported that Google parent Alphabet agreed to pay $700 mn and make certain changes to its app store in settling an antitrust challenge to the company. The settlement resolves claims by a group of states that Google operated its app store, Google Play, as an illegal monopoly, allegedly curbing competition from other app distributors on devices using the Google-owned Android operating system.

Developers will now be able to use an alternative billing system to Google Play’s billing option, which the company said it has been piloting for more than a year. The settlement also requires that Alphabet simplify the process of downloading apps directly from developers’ websites rather than using an online store such as Google Play.

Alphabet said its operating system and app store give more choices to consumers than their competitors, and the company is committed to improving Android and Google Play. ‘We’re pleased to reach an agreement that builds on that foundation,’ the company said.

Ben Maiden

Ben Maiden is the editor-at-large of Governance Intelligence, an IR Media publication, having joined the company in December 2016. He is based in New York. Ben was previously managing editor of Compliance Reporter, covering regulatory and compliance...