OpenAI fired Sam Altman, its CEO-founder, the afternoon of November 17, 2023. In the corporate chaos that followed the company went through two acting CEOs within three days. By the end of the day on Monday, November 20, Altman and other key executives at OpenAI were hired by Microsoft. Immediately following this 700 employees – practically the whole company – signed a letter threatening to quit unless the Board resigns and Altman is reinstated.
It’s hard to imagine how this termination could have been handled more ineffectively. It is an exemplar of exactly how NOT to fire your CEO.
However, all the OpenAI Board members had to do was ask their own product – ChatGPT – how to go about terminating Altman. Here’s my query to ChatGPT-4, and it’s response –
PROMPT TO GPT:
I am on the Board of Directors of a closely held California-based corporation. We are considering terminating our CEO. Please tell me in
To provide better governance and ensure rational decision-making, GPT-5 will join the new OpenAI Board of Directors
detail what steps we should take to prepare for and execute this termination. Take into consideration that my company is very high profile – we are a cutting-edge technology company that receives a lot of attention from the press and the technology industry. Our CEO is one of the founders of the company, and he is effectively “the face” of the company. He is frequently interviewed by the industry press and often appears at industry conferences and events to talk about our products. We have many partnerships with other companies and the CEO has been instrumental in forming those partnerships. While we are a bit unsure about this, we think that many of our key employees respect the CEO and may be loyal to him
CHATGPT-4 RESPONSE (my comments in italics and [brackets]):
Given the high-profile nature of your company and the CEO’s prominent role as a founder and the face of the company, the process of terminating the CEO requires additional layers of consideration and caution. Here are detailed steps tailored to your situation:
Preliminary Confidential Discussions: Initiate confidential discussions among the Board of Directors to thoroughly assess the reasons for considering termination and the potential impact on the company. Given the CEO’s prominence, it’s essential to evaluate the risks and benefits of this action carefully. [OpenAI board appears not to have done this]
Legal Consultation: Engage legal counsel experienced in high-profile executive terminations and corporate law. They can provide advice on contractual obligations, potential legal risks, and strategies for minimizing legal exposure, especially if the CEO is also a shareholder or has a significant stake in the company. [No information on whether Altman has an employment agreement with OpenAI or what its terms are. Note, however, that noncompete restrictions are not enforceable under California law, and therefore Altman and the other OpenAI employees cannot be prevented from going to work at Microsoft for this reason. There may, however, be trade secret issues in this fact pattern
Review Employment Agreement and Corporate Bylaws: Scrutinize the CEO’s employment contract and the corporation’s bylaws. These
Employee Letter to OpenAI Board
documents will guide the legal and procedural aspects of the termination process, including any ‘for cause’ conditions, severance terms, and required board approvals. Conduct a risk assessment focusing on potential repercussions within the technology industry, impact on partnerships, employee morale, and public perception. This assessment should consider the possibility of the CEO’s departure affecting key relationships and projects. [The OpenAI board either failed to do this, or completely miscalculated]
Internal Alignment: Ensure alignment and consensus within the Board of Directors. Given the CEO’s foundational role in the company, board members must be unified in their decision and prepared for the subsequent steps. [Ilya Sutskever, one of the Board members that voted to terminate Altman on Nov. 17, signed the employee letter on Monday Nov. 20]
Plan for Succession and Transition: Develop a succession plan that includes identifying an interim CEO or a successor. This plan should address how to maintain continuity in leadership and reassure employees, partners, and stakeholders about the company’s stability and future direction. [OpenAI announced an interim CEO on Friday – Mira Murati – and on Monday replaced her with a person brought in from outside the company. Over a period of four days OpenAI had three CEOs]
Communications Strategy: Formulate a comprehensive communications strategy. This should include internal communication to employees, as well as external communication to the press, industry partners, and stakeholders. The message should be crafted to maintain the company’s reputation and market confidence. [Internal/external communications were almost nonexistent]
Managing Employee Sentiment: Prepare for potential fallout among employees who may be loyal to the CEO. This could involve individual discussions with key team members, departmental meetings, and ensuring that HR is equipped to handle any concerns or unrest. [Seems not to have been done at all]
Board Approval and Documentation: Formalize the decision with a board vote and document the process meticulously. This documentation is crucial for legal compliance and for maintaining a record of the decision-making process. [Outside of the board, no one seems to know why Altman was fired]
Executing the Termination: Plan the termination meeting with the CEO carefully, considering the timing, location, and who will be present. The conversation should be handled with respect and sensitivity, acknowledging the CEO’s contributions to the company. [Termination was done via Google Meet, a video service similar to Zoom]
Public Announcement: Coordinate the public announcement of the CEO’s departure with the communications strategy. This announcement should be respectful and align with the narrative agreed upon by the board. [The board didn’t have a narrative]
Post-Termination Steps: Monitor the company’s internal and external environment after the termination. Be prepared to manage any unforeseen consequences, including media inquiries, market reactions, and internal dynamics.
After the termination, review the company’s policies, leadership structure, and strategic direction. This is an opportunity to learn from the experience and strengthen the company’s governance and leadership practices.
Throughout this process, the key is to balance legal and procedural diligence with sensitivity to the human and cultural aspects of this significant change. The termination of a high-profile CEO can have far-reaching implications, and managing it effectively requires a thoughtful, strategic approach.
Earlier this year, on the eve of trial in Baker v. Goldman Sachs in federal district court in Boston, I published a blog post describing the facts behind this unusual case, which involved the acquisition of Dragon Systems by Lernout & Hauspie in a $600 million all-stock deal. Soon after the acquisition closed the market discovered that Lernout had fabricated its Asian sales figures. This was quickly followed by Lernout’s bankruptcy, which left Dragon (owned by the Bakers, husband and wife founders) holding worthless Lernout stock. (Baker v. Goldman Sachs – The Business Deal From Hell).
The acquisition was negotiated and concluded in the first half of 2000, just as the technology bubble was beginning to deflate.
After a lengthy trial the jury ruled in favor of Goldman Sachs on all issues except the claim that Goldman violated M.G.L. c. 93A, the Massachusetts statute that makes illegal “unfair or deceptive acts or practices.” Under Massachusetts law, that claim must be decided by the judge.
Now, Massachusetts federal district court judge Patti Saris has issued her decision on the Baker’s 93A claims, holding that Goldman Sachs did not violate 93A. This ruling is not a surprise; judges rarely find a violation of 93A when a jury rules against a plaintiff on the underlying claims, which in this case were negligence, breach of fiduciary duty and fraud.
However, her opinion is a fascinating look into how a transaction of this magnitude can go wrong. In addition, I read her opinion to imply that, in her view, the jury should have found Goldman Sachs negligent. Of course, there is no explanation for why the jury ruled against the Bakers – perhaps the Bakers were unsympathetic witnesses, or they drew a hostile jury. Perhaps the jury failed to understand the case, or the Baker’s case was poorly tried. These are the risks of the jury system–you may be convinced you have a great case and the jury can still rule against you. And, to rub salt in your wounds, you will never know why.
But, the case shows what goes on behind the scenes in a transaction of this sort, especially when your investment banker is Goldman Sachs, which was paid $5 million for its services to Dragon Systems. Here are a few points from Judge Saris’ opinion that caught my attention:
The Goldman Sachs Investment Banking Team Was Very Young. The team was comprised of three people. The leader of the team was only 31 years old. The second team member was 25 years old and was job hunting at the time. The third investment banker was a 21 year old recent college graduate. No senior Goldman Sachs investment banker did any work on the transaction.
=> The fact that Dragon allowed a 31 year old Goldman employee to be the leader on a $600 million deal of this importance to Dragon’s owners is a mystery. This brings to mind the saying, “young doctors, old lawyers.” For this case, I would change that to read, “young doctors, old lawyers and old investment bankers.” A degree from a prestigious business school does not equal experience.
Dragon Made a Critical Decision Without the Involvement of Its Investment Bankers. The sale of Dragon to Lernout initially had a large cash component – 50% cash/50% Lernout stock. But, this changed at a meeting at which Goldman Sachs was not present. Judge Saris wrote that “Janet Baker chose to pursue an all-stock transaction with [Lernout] for $580 million. By the end of the meeting, Janet Baker and [Lernout] signed a handwritten agreement setting forth a fixed exchange ratio for an all-stock acquisition of Dragon. … Janet Baker made this napkin agreement without consulting Goldman.”
=> The fact that the Bakers modified the deal without consulting Goldman Sachs suggests a lack of business sophistication on the part of the Bakers. A “napkin agreement” to sell your company for 100% stock, without including your investment banker in the decision? This was a huge decision, and by the time it was made it was too late for Goldman Sachs to advise against it. If you’re going to pay your New York investment bankers $5 million, keep them in the loop.
Goldman Sachs C.E.O. Lloyd Blankfein and C.O.O. Gary Cohn, in the boardroom of Goldman’s headquarters, in New York City
The Investment Bankers Are Not Your Friend, So Get Their Opinions and Assurances in Writing. Judge Saris wrote that “Goldman was dissatisfied with respect to Lernout’s answers on many due diligence questions up until the last moment. The transaction would likely not have gone forward if Goldman had voiced its ongoing concerns about financial due diligence. In his deposition, Wayner said he did not disclose any concerns he had about due diligence at the March 27 meeting because ‘the client did not ask.’”
=> Wow. Wayner was the senior Goldman team member. His testimony that he was concerned about due diligence but didn’t voice his concerns because “the client did not ask” shows how careful a client must be in dealing with its investment banker. When you are about to sell your company for $600 million in stock you should be waking up in the middle of the night worrying. And, before you fall back to sleep you should be emailing your investment banker and asking, “is there anything about this deal you’re concerned about that we should know before we jump?” The fact that an investment banker as prestigious, expensive and (supposedly) competent as Goldman Sachs took a “don’t ask, don’t tell” position toward its client says a lot about where investment banker interests lie. Could their $5 million fee on the sale of the company been in the back of Wayner’s mind when he weighed whether to tell the Bakers about his concerns?
Don’t Assume the Investment Banker is Creating a File You Can Use to Sue The If Things Go Wrong. Judge Saris wrote that Goldman Sachs has a “written policy encouraging its mergers and acquisitions department not to safeguard their written notes. The policy tells investment bankers to keep ‘[n]otes supporting due diligence,’ to throw out ‘[n]otes to self-citing unresolved problems,’ and ‘[w]hen in doubt, throw them out,’ unless litigation has commenced.”
=> In case you had any doubts, that’s what’s going on behind the scenes at Goldman Sachs, and probably other New York investment bankers. If you make a note to yourself regarding an unresolved problem in an M&A transaction, throw it out. Why? So it can’t be used in litigation against Goldman Sachs if there is a legal problem later.
Near the end of her lengthy opinion Judge Saris observes that “The reasons why small startup companies like Dragon go to a place like Goldman to assist with hatching their golden eggs is because they don’t have their own expertise to analyze revenue projections by asking tough questions to potential merger partners.” Reading between the lines of her opinion, one could conclude that Judge Saris felts that Goldman Sachs did not provide that expertise to Dragon. However, this decision was the province of the jury, and the jury ruled for Goldman Sachs on all of Dragon’s legal claims. Judge Saris’ ruling on Dragon’s Chapter 93A claims was Dragon’s last chance, and she concluded that Goldman Sachs did not engage in conduct that met the test of “unfair and deceptive practices,” as that law has been interpreted by the courts. Absent an appeal, her decision marks the end of a long road, strewn with missteps by Dragon and its advisors. Perhaps it will serve as a lesson for the future.
An interesting “David v. Goliath” jury trial is scheduled to begin in Massachusetts U.S. District Court Judge Patti Saris’s Boston courtroom this week. The case has received a fair amount of press coverage, but not nearly enough in my opinion. (Steven Syre Boston Globe column today, July 2012 NYT article).
The events in Baker v. Goldman Sachs date back to the heady days of the dotcom era. In short, James and Janet Baker (pictured here) spent much of their careers pioneering speech recognition technology which they commercialized under their company Dragon Systems, based in Newton, Massachusetts. The Bakers were legendary in the Massachusetts tech community in the 1990s – home-based technologists who came up with a promising-today, sky-is-the-limit-tomorrow technology.
In 2000 they sold Dragon for almost $600 million to Lernout & Hauspie, a Dutch company. Unfortunately, they were paid fully in Lernout stock, and almost immediately after the sale Lernout was discovered to have been cooking its books. The stock went to zero, and so did the consideration the Bakers received for their company. For the Bakers, this was a business catastrophe of mythical proportions. One day they owned an immensely valuable company that that owned technology they had spent decades developing. The next day they had sold the company in exchange for almost $600 million of Lernout stock. Within a few months, their company was gone and their stock worth nothing.
Goldman Sachs was the “investment banker” for Dragon’s sale to Lerner, and the question is, why didn’t they discovery that Lernout was a shell game? The gist of the Baker’s case is that Goldman (which received $5 million in fees) breached various legal obligations to Dragon by failing to discover that Lerner was “all hat, no cattle.” According to Steven Syre’s column, after Dragon’s salel to Lernout The Wall Street Journal simply called purported L&H customers and discovered they weren’t doing business with Lernout, something the Bakers claim that Goldman should have uncovered as part of its due diligence before the sale. In fact, it appears that Lernout’s Asian business—a significant part of its overall revenues—was a sham. Lernout had fabricated almost $400 million in income.
To make matters worse, the Times article asserts that Goldman assigned a young (20s and 30s) and inexperienced team to handle the deal. The Bakers’ court filings claim that Dragon and the Bakers fell victim to a massive fraud, that there were many “red flags” that Goldman failed to pursue in due diligence, and that had Goldman done the job it promised to do, the sale never would have occurred. In fact, the Bakers claim that Goldman had discovered problems with Lernout on behalf of another client, and failed to disclose these problems, or raise its level of investigation based upon this knowledge.
The full story is complex, and involves many different actors at Dragon, Goldman and various accounting and law firms. The question of who should have discovered Lernout’s fraud involves more finger-pointing than the presto movement in a piano concerto for four-hands. Of course, the Bakers (local scientists with significant bona fides) are far more sympathetic than the Goldman witnesses, many of whom left Goldman shortly after the sale to Lernout closed in June 2000. It remains to be seen whether they can hold their own as trial witnesses. Goldman Sachs’ credibility? Well, the Bakers must hope they draw a jury that reads the business pages.
While in the eyes of a non-lawyer the facts (as alleged) appear to favor the Bakers, Goldman has a number of technical legal defenses, and while it’s had little success with them before Judge Saris, presumably it hopes to prevail on them on appeal. For example, it argues that it was not responsible for the due diligence that would have uncovered the irregularities in Lerner’s sales numbers — Dragon’s accountants were responsible for this task. It claims that Lerner’s fraud was essentially undiscoverable by Goldman. It’s trump card is that the Bakers are caught in a legal catch-22: since Goldman Sachs’ engagement letter was with Dragon (the corporation), the Bakers lack legal standing as individual shareholders. And, since Dragon was merged into Lernout (and therefore for reasons arising out of bankruptcy law had no right to sue), Goldman is immune from liability to Dragon. Since neither the Bakers nor Dragon has standing, Goldman has no liability. Thus far the Bakers have been able to side-step this argument (arguing, for example, that they are third-party beneficiaries to the contract), but this issue remains unresolved, and could prove to be a get-out-of-jail-free card for Goldman.
The parties’ joint pretrial memo, outlining the many legal and factual issues, is linked below.
It is surprising that this case has not yet settled (the vast majority of civil cases—estimated as high as 95%—do). However, cases often settle “on the courthouse steps” or during trial. In fact, the trial judge in this case has been known to require party principals to hold a settlement conference following opening statements at trial. The idea is that once the parties get a look at a real jury, and hear the opposing attorney’s opening statement, they may adjust their expectations and reach a settlement.
If this case doesn’t settle it is almost certain to raise interesting and important investment banking legal issues both during trial, and on appeal to the First Circuit. Presumably, Goldman Sachs has revised its engagement letters.
Update: The jury found against the Bakers at trial.
Oriental Financial Group, Inc. v. Cooperativa De Ahorro y Crédito Oriental (1st Cir. October 18, 2012) — In this case the First Circuit adopts the trademark law “progressive encroachment doctrine,” joining the 6th, 7th, 8th, 9th and 11th circuits. The progressive encroachment doctrine may be used as an offensive countermeasure to the affirmative defense of laches (delay in brining suit) where the trademark owner can show that “(1) during the period of the delay the plaintiff could reasonably conclude that it should not bring suit to challenge the allegedly infringing activity; (2) the defendant materially altered its infringing activities; and (3) suit was not unreasonably delayed after the alteration in infringing activity” (quoting Oriental Financial).
Harlan Laboratories, Inc. v. Gerald Campbell (D. Mass. October 25, 2012) — Applying Indiana law, Judge Patti Saris issues a preliminary injunction enforcing a one year non-compete agreement. However, the opinion makes liberal use of Massachusetts and First Circuit precedents.
Blake v. Professional Coin Grading Service (D. Mass. October 6, 2012) — In this case, which involves alleged trade secrets associated with a method to grade the “eye appeal” of coins, Judge William Young concluded that the “method” was not subject to trade secret protection due to the fact it had been publicly disseminated before being disclosed to the defendants. However, Judge Young ruled that the case could proceed based on the alleged misappropriation of a proposed marketing plan. In addition to his analysis of trade secret law, the case contains an extensive discussion of Lanham Act issues including “reverse confusion” (which is always confusing) as well as the application of Massachusetts law to the intellectual property issues raised in the case (conversion, breach of contract, the covenant of good faith and fair dealing, unjust enrichment and civil conspiracy).
In DeJesus v. Bertsch, Inc. (D. Mass. Oct. 16, 2012) Judge Young conducts a detailed analysis of corporate successor tort liability under the Massachusetts “de facto merger” and “mere continuation” exceptions. In this case he concludes that the defendant corporation is not subject to successor liability.
While I shy away from posting PowerPoint outlines on this blog, the materials from two talks that my partner Sean Gilligan recently gave to attorneys in our firm are sufficiently comprehensive as to be an exception. Both outlines are on scribd.com, and are embedded below:
In this time of economic uncertainty, what can a founder of a startup do to increase his chances of attracting an outside investment and maximize the likelihood of his ultimate financial success? The answer may be counterintuitive: founders should not try to “hedge” their commitment to their new business in an effort to minimize downside risk. Without the founders taking on some risk, making sacrifices and giving an unfettered commitment to their startup, they will have a much harder time attracting investors and achieving their ultimate goals. Continue reading ….