Written as part of this project on reforms in EA.
One theme that came up a lot in discussions of possible changes in EA was the idea of better support for whistleblowers or other people raising problems. We put together some information and ideas on this area.
Raising concerns
In any organization, it’s vital that problems can be understood and addressed. These might include
- A lack of good policies or practices
- Policies exist but are not being followed
- Bad working conditions or unfair treatment of staff
- Cultural or interpersonal problems within the organization
- Disagreement about strategy
Whistleblowing
The term whistleblowing is typically used for more serious problems, such as
- Breaking the law
- Risks to public health or safety
Government protections for whistleblowers are typically limited to specific types of problems — for example the UK defines it as relating to wrongdoing that affects the public interest (rather than workplace disputes that don’t affect the general public).
Some things that seem good
- Organizations should have an official whistleblower policy shared with staff, saying that staff will not be punished for making good-faith reports to management or to the appropriate government agency. One template can be found on the Anti Entropy resource portal.
- Organizations should actually follow the spirit of that policy — meaning that there’s neither formal nor informal retaliation for good-faith reports.
- The effective altruism ecosystem should uphold the above; retaliating against people who make good-faith reports should be bad for an organization’s reputation in EA. (But how to operationalize this is complicated.)
Options for reporting problems
Here are some options for escalating problems, though many of these won’t be suitable for a given situation. Some are more suitable for less serious internal problems, and others for more serious problems with repercussions beyond the organization. Some of these likely violate organizational policies if you work at the organization; please see legal resources below.
- You could report the problem within the organization.
- This might look like talking to HR staff, finance staff, or a different part of management.
- Talking to other lower-level staff can also be helpful to get a clearer picture of the problem.
- Note that one person I talked to who had investigated corruption in non-EA organizations said that in some cases a crooked department or organization will be glad if you report, because they’ll know who to fire and the corruption can proceed more smoothly.
- Even in less crooked organizations, you should consider that you may experience retaliation.
- You could contact someone on the board of the organization, or write to the entire board.
- You could contact a funder of the organization. For example, Open Philanthropy offers several ways to contact them, including about concerns about their grantees.
- You could talk to services that recommend the organization. For example the 80,000 Hours job board sometimes decides not to list job openings at organizations they’ve heard concerns about. Or the team that puts on EA Global may not want to give the organization a booth at the organization fair.
- You could talk to the community health team at CEA. (Julia, the main author of this piece, works there.) Their ability to help may be limited by practical and legal considerations.
- If something illegal is happening, or if there’s a health / safety problem, you could report it to the relevant government agency.
- You could write about the problem publicly. This could be under your own name, could be anonymous, or could be a group piece from several people familiar with the problems.
- You could talk to other people might who write about it (journalists, or projects like Omega).
Legal protections
The UK has relatively straightforward protections for workers reporting certain problems to their employer or to the relevant government agency. The US has a more complicated and patchy set of protections, varying by what problem you report and by state. See more information in the “Resources” section.
Things that are not protected by any whistleblowing laws I looked at, if I understand right:
- Reporting problems beyond the specific types spelled out in the law
- Reporting problems about entities you don’t work for (because the laws are set up to protect you from your own employer)
- Complaints by people who aren’t an “employee” or “worker,” e.g. an intern or volunteer
- Publicly disclosing a problem or telling the media
- Your ability to work for other employers in the field
- Your chances of getting grants (except maybe if you worked for the grantmaker)
- Your reputation in the community
Resources on whistleblowing
UK:
US:
Germany:
What does Germany’s new Whistleblower Act mean for employees?
Canada:
Whistleblowing Canada
Australia:
Financial support for people reporting problems
One possibility that came up in discussion over the last year was the possibility of financial support for whistleblowers in EA. The US government offers some financial rewards for information that leads to enforcement on certain types of fraud or financial crimes. The SEC routinely makes multi-million-dollar awards to whistleblowers who reveal financial fraud.
But there’s a wide range of harmful behavior that EAs care about, outside of financial fraud. It’s much less clear how you would run a good rewards program for this wider range of problems, or what the eligibility should be.
In an EA-adjacent space, there was a temporary offer of financial reward for information about an organization that some people were interested to know more about. In another situation, a person writing up problems at an organization paid two former staff members for their efforts in raising the problems.
I think there’s also the possibility of a more informal network of people supporting each other in whistleblowing situations, for example if they know a friend is considering leaving a job in a bad work environment, or if a friend or coworker has been fired. People may find it easier to assess specific situations than to precommit to rewards in situations that haven’t happened yet.
Whistleblowing in AI safety
This post doesn't aim to cover whistleblowing about harmful practices at AI labs.
Julia - I appreciate this initiative, and just want to add a caveat.
I think with any policies and procedures for 'reporting concerns' or whistleblowing, it's important, as in any 'signal detection problem', to balance the risks and costs of false positives (e.g. false accusations, slander from disgruntled or mentally ill employees) against the risks and costs of false negatives (missing bad behavior or bad organizations).
My impression is that EA has suffered some important and salient false negatives (e.g. missing SBF's apparent sociopathy & FTX frauds). But some EA individuals and organizations, arguably, have also been subject to a wide range of false allegations -- especially by certain individuals who have a very long history of false allegations against many former associates and former employers.
It can be very easy to be taken in by a plausible, distressed, emotionally intense whistleblower - especially if one has little professional experience of handling HR-type disputes, or little training in relevant behavioral sciences (e.g. psychiatry, clinical psychology). This is an especially acute danger if the whistleblower has any of the Cluster B personality disorders (antisocial, narcissistic, borderline, histrionic disorders) that tend to be associated with multi-year histories of false allegations against multiple targets.
And these problems may be exacerbated if there are financial incentives for making false allegations (e.g. 'financial support for people reporting problems'), without many social or professional costs of doing so (e.g. if the false allegations are made from behind a cloak of anonymity, and their falseness is never reported to the EA community).
Thus, I would urge any EAs who set themselves up as adjudicators of whistleblowing cases to get some serious training in recognizing some of the red flags that may indicate false allegations -- especially in assessing any patterns of persistent false accusations, mental illness, or personality disorders.
It only takes one or two people with serious borderline personality disorder (for example), who are willing to make multiple false allegations, to ruin the reputations of multiple individuals and organizations -- especially if the people trying to investigate those allegations are too naive about what might be going on. The same caveat applies to any EAs who take it upon themselves to do any independent 'investigative reporting' of allegations against individuals or organizations.