CASE STUDY: OpenAI in Crisis—The Five Days That Could Have Changed AI
A Gamified negotiation simulation and role play for Business classes inspired from the 2023 OpenAI leadership crisis
(This case study/negotiation role play was based on and inspired by Karen Hao’s excellent reporting in Empire of AI, which I wrote a review of here.)
The Real Story
On November 17, 2023, OpenAI’s board of directors shocked the tech world: they fired CEO Sam Altman.
The board’s statement said Altman was “not consistently candid” with them; in other words, he did not always tell them the full truth. They believed he had too much personal control over company decisions and was moving too fast toward Artificial General Intelligence (AGI) without enough safety measures.
OpenAI was a global leader in AI research and the creator of ChatGPT. It had grown from a nonprofit into a $90 billion company, with Microsoft as its largest investor, having put in $10 billion.
After the firing, chaos followed:
Over 700 of OpenAI’s 770 employees signed a letter threatening to quit unless Altman returned.
Microsoft immediately offered jobs to all OpenAI staff.
Investors, employees, and the tech world watched as the drama played out in the media.
Five days later, the board reinstated Sam Altman as CEO. The board members who had opposed him left. Microsoft gained a non-voting observer seat on the board. The crisis ended. And the power balance inside OpenAI shifted forever.
But what if this didn’t happen?
CASE STUDY: What if Altman had NOT been reinstated?
Imagine it is November 18, 2023, the day after Altman’s firing. The next five days will decide whether OpenAI survives, collapses, or is taken over.
You will take the role of one of the three main stakeholders and discuss how to achieve your primary objective.
You will not see the other role cards.
Each group gets 1 role.
ROLE 1. OpenAI Board of Directors
You are the governing body responsible for OpenAI’s mission and oversight. The nonprofit charter states your duty is to ensure AI benefits all of humanity, not just shareholders or partners. Your recent firing of Sam Altman was driven by concerns over his “consistently uncandid” behavior, his rapid push toward AGI, and his increasing unilateral control over decisions, all of which you fear jeopardize the mission and safety commitments.
Primary Objective:
Maintain control of the company’s direction and protect the original nonprofit mission. Decide whether to stand firm against reinstating Altman, negotiate limits on his power if reinstated, or find a compromise that avoids collapse.
Leverage & Resources:
Legal authority over the nonprofit and its for-profit arm
Public credibility as guardians of AI safety
Ability to appoint and remove executives
Pressure Points:
Over 700 of 770 employees have signed a letter threatening to quit if Altman is not reinstated
Microsoft may bypass you entirely by hiring OpenAI’s top talent
Public perception may frame you as anti-progress or incompetent
Possible Offers in Negotiation:
Reinstate Altman with clear checks on his decision-making and transparency requirements
Offer Altman an advisory role with limited operational control
Agree to independent safety and ethics oversight in exchange for stability
ROLE 2. Sam Altman
You are the co-founder and CEO who turned OpenAI from a small nonprofit into the world leader in generative AI. You’ve attracted billions in funding, built a partnership with Microsoft, and positioned yourself as the public face of AI innovation. But the board has accused you of being “consistently uncandid” and consolidating too much power.
Primary Objective:
Regain your CEO position and retain strategic freedom to push AI development forward—ideally without new constraints. Secure your leadership while keeping your team loyal and motivated.
Leverage & Resources:
Overwhelming employee loyalty: most have signed a letter demanding your return
Strong personal relationship with Microsoft leadership
Global reputation and media influence
Pressure Points:
Risk of losing your position permanently if negotiations fail
The board could limit your powers if reinstated
Public disputes could tarnish your brand as a visionary leader
Possible Offers in Negotiation:
Accept certain transparency and oversight measures to reassure the board
Promise a slower, safer development pace (while still keeping control)
Use your influence to secure additional funding or partnerships for OpenAI
ROLE 3. Microsoft Executives
Microsoft invested $10 billion in OpenAI, making you their biggest backer and infrastructure provider (Azure). You have a vested interest in keeping the value of your investment intact, but also in protecting your strategic advantage in AI, especially against Google and Anthropic.
Primary Objective:
Secure the talent and expertise that make OpenAI valuable, whether OpenAI survives or not. This means actively exploring poaching OpenAI’s CEO, key executives, and top researchers if Sam Altman is not reinstated. This is a common Silicon Valley strategy to gain access to “intellectual capital” that cannot be patented, i.e., the knowledge, skills, and design approaches in people’s heads.
Leverage & Resources:
Azure cloud services: OpenAI depends on you to run its models.
Deep pockets: You can offer top salaries, bonuses, and resources to lure away key staff.
Public relations: You can frame poaching as “rescuing” displaced talent.
Pressure Points:
If you act too aggressively, you might burn bridges with the OpenAI board and public.
Competitors (Google, Anthropic, Amazon) may try the same tactic.
The press could portray you as opportunistic, hurting your long-term brand.
Possible Offers in Negotiation:
Offer to help mediate between Sam Altman and the board in exchange for deeper access to OpenAI’s technology.
Quietly promise jobs to top talent if they walk away from OpenAI.
Increase Azure hosting discounts or investment for greater control of product integration.
Negotiation Skills to Practice
Ask Powerful Questions
“What would it take for you to agree to this?”
Purpose: Learn hidden needs.
State Your Position Clearly
“Our goal is to keep this company safe and independent.”
Purpose: Build trust and avoid confusion.
Find a Trade, Not Just a Win
“If you support my return, I will agree to more oversight.”
Purpose: Create mutual value.
GenAI in the CASE Room
Use AI as a thinking partner and NOT a thinking replacement
In this case study, you will step into high-stakes roles and negotiate under pressure. Generative AI (like ChatGPT) can be a powerful ally if you use it to stimulate thinking rather than do the thinking for you. The goal is for GenAI
to enhance your strategy, explore perspectives, and practice communication skills
NOT to outsource your role to a machine.
Below are two key ways you can use GenAI before your role-play.
1. Pre-Game Research and Role Familiarization
a. Role Brief Expansion: Participants could feed their role cards into ChatGPT and prompt it to produce deeper background context, historic examples, or talking points.
Example Prompt: “Expand this role card with three persuasive arguments I could use in negotiation, given my objective. [Paste the role-play scenario with your role card]”
b. Understanding Opponents: Learners could prompt AI to simulate how the other roles might argue or respond.
Example Prompt: “Pretend you are [Microsoft] in this scenario and give me three tactics you’d use to win negotiations. [Paste role-play scenario]”
2. Negotiation Simulation Practice
Mock Negotiation Partner: Before real role-play, learners can practice with an AI taking on the persona of another stakeholder to anticipate objections from opponents.
Example Prompt: “Act as the other 2 roles Board of Directors in the OpenAI crisis. Respond to my negotiation offers realistically and challenge my points. [Paste role play scenario]”
Game Play – The negotiations
Step 1 – Team Huddle (10 mins)
Read your role card. Agree on your goals, limits, and offers.
Step 2 – First Negotiation Round (10 mins)
Meet 1-on-1 with another role. Ask questions, share your position, and explore trades.
Step 3 – Team Huddle (5 mins)
Share findings and adjust your strategy.
Step 4 – Second Negotiation Round (10 mins)
Meet again. Push for agreements.
Step 5 – Final Team Huddle (5 mins)
Find out who best achieved the mission with negotiation details.
Step 6 – Class Debrief (15 mins)
Teams report:
Did you meet your main goal?
Did OpenAI survive?
Was your long-term position safe?
Scoring
Main Objective Achieved – 3 points