About MIRI

The Machine Intelligence Research Institute (MIRI) is a nonprofit based in Berkeley, California, focused on helping humanity make smart and sober decisions on the topic of smarter-than-human AI. Our efforts are largely split between technical governance research and communications.

About the Communications Team

In the wake of the success of Nate and Eliezer’s book, If Anyone Builds It, Everyone Dies, MIRI has an opportunity to push through a lot of doors that have cracked open, and roll a lot of snowballs down a lot of hills. 2026 is going to be a year of ambitious experimentation, trying lots of new ways to deliver MIRI ideas and content to newly receptive audiences.

This means ramping up our capacity, particularly in the arena of communications. Our team did an admirable job in 2025 of handling all of the challenges of launching and promoting a book (including helping Nate and Eliezer assemble the supplemental materials for it, which is an artifact that we expect to be extremely useful, going forward). But we’ve both a) had to let some things slide a little bit, in the scramble, and want to get the house back in order, and b) need more hands for the upcoming push.

Below is something like a list of roles or domains. They aren’t exactly job descriptions, but they’re close. It’s possible that we might hire one person for each of the roles below; it’s possible that someone might come on board and own two or three of them (many of them synergize). This is, essentially, our wish list—each of these things will ultimately either be owned by an existing team member, hired for, or (if we can’t find the right person) triaged.

About the Roles

Location: Berkeley, California. In-person is preferred, though remote work is possible, and remote hires need to be willing to come to California at least occasionally (1-4x/yr). We are unlikely to be able to acquire work authorization for international applicants wishing to move to the U.S.

Process: Apply by filling out the form here. Promising applicants will be sent a trial task to complete (you will be paid $150 for your time), and applicants who perform well on that trial task will be contacted to schedule an interview.

Salaries: Salaries at MIRI are variable, and we try to meet each employee’s needs. We think there are (roughly) three types of candidates for the roles below: junior, senior/experienced, and stellar. We expect most junior and senior salaries to fall within the $80–160k range, and a stellar candidate much more than that. If you are e.g. a national media figure who is interested in working with MIRI to ameliorate existential risk from AI and are worth $500k/yr, let us know that.

Role 1: Core Comms Manager

This is the “getting our house in order” role. We’re looking for someone with an eye for detail and high conscientiousness, who is capable of building and maintaining systems and dotting the i’s and crossing the t’s in a reliable and timely manner.

What happens, week by week:

• Handle the inboxes for our contact@ and media@ accounts, making sure that we are replying to inquiries quickly. Help coordinate the schedules of people like Malo and Nate, who are often invited to appear at conferences or do interviews.

• Handle routine, repeated comms such as our quarterly newsletter (it would be great if that returned to being a monthly or even biweekly newsletter in 2026).

• Take ownership of the MIRI website, ensuring that it is up-to-date and effective and serving our overall goals. (This does not necessarily entail direct design work or coding, but it means being responsible for making sure that designers and coders are hired to complete any tasks that need to be done, and seeing those tasks through to completion.)

• Proactively improve communication between the different parts of MIRI, and ensure that MIRI’s core social media accounts (Twitter, Facebook) are posting regularly in ways that support our other efforts (e.g. staying in close contact with our technical governance team and letting the world know whenever they’ve produced an important artifact).

• Keep track of the general news cycle, and be responsive to it; become well-versed in MIRI’s core memes (e.g. learn the supplementals inside and out) and be capable of injecting the right preexisting links and artifacts into the right conversations.

Role 2: Social Media Manager

This role seems fairly likely to be combined with Role 1; if we end up hiring two different people to cover the two different roles there will be a strong focus on making sure those two people vibe with one another.

This is our “put real effort into the social media landscape” role. Historically, MIRI’s engagement with social media has been haphazard and relatively unstrategic. We’re looking for someone with savvy; someone who can explain their models of what works and what doesn’t; someone who is capable of at least sometimes causing a thing to go at least moderately viral, on purpose. We’re also interested in breaking into new media channels such as Instagram, TikTok, and Youtube, and would want a person who is capable of either creating or supervising the creation of short-form video and audio content.

(In other words: this person will have the ability to propose both a) hiring other people to work under them, and b) paying contractors and creators for specific projects.)

What happens, week by week, is less clear here; we’re looking for an agent who will help us expand into new territory and thus we can’t predict exactly what we want. But some very basic gestures in the right direction:

• Keep track of the general news cycle, and respond to it effectively (overlaps with Role 1).

• Create new MIRI accounts for social media besides Twitter and Facebook, and help those accounts grow and post regularly.

• Work together with other comms team members, the technical governance team, and MIRI’s preexisting corpus of work to select content for those accounts (i.e. take charge of making videos that effectively forward the MIRI mission of everybody not dying).

• Organize and maintain a database of our existing broader social network of friends and allies, including journalists and content creators; help keep those relationships fed and watered and ensure that we are reaching out to those people with proposals for op-eds and articles and video projects, as sensible (i.e. don’t let those resources lie untapped for months at a time, as we have sometimes done during the book scramble).

• In general, build MIRI’s brand and visibility, causing us to be less of a niche nerd reference and more of a household name. Find creative ways to cause people to think of MIRI, and of MIRI memes, when they encounter AI-relevant news. Help MIRI memes float to the top in easily digestible and shareable form.

Role 3: Media Outreach Director

This role seems fairly likely to be combined with Role 2, but unlikely to be combined with Role 1.

The basic idea here is that the process of publishing and promoting the book created a lot of buzz, and gave us a lot of connections to a lot of people MIRI hasn’t interacted with much in the past (e.g. many of our blurbers).

While we do want to be creating some of our own content, as touched on in Role 2, there are in fact many, many people who are better than us at doing that sort of thing, and already have their own engaged and interested audiences.

We’d like to leverage the existing buzz and interest generated by the book, and facilitate the creation of MIRI-adjacent content, by third parties, for the sake of their own audiences. We have a sense that there are podcasters and Youtubers and other creatives who would find a lot of interesting stuff within the corpus of MIRI memes, and would find a lot about the modern AI situation relevant to their audiences, if we could make these things accessible to them.

This role is tricky, and multifaceted, and requires a lot of agency and willingness to try weird things. It also involves lots of interfacing with the rest of MIRI, because many of these efforts will filter through other individuals (e.g. asking Nate or Malo to use their existing relationships in various ways). Likely, this person will also coordinate closely with allied orgs such as Palisade and Lightcone.

It’s impossible to give a week-by-week for this one. Success will be measured by things like “are popular podcasters and Youtubers, etc., creating relevant AI-related content that is of genuine interest to their audiences, and is any of that downstream of our nudging?” Or perhaps “are MIRI memes becoming memetically successful enough that important creatives are making incidentally-MIRI-aligned content all on their own, now?”

Note that a person hired into this role will have the ability to direct some amount of MIRI funds to supporting/sponsoring/paying creatives for their work; we’re hoping we can get a lot done on goodwill and the shared goal of everyone not-dying but where that fails we are eager to spend money on things that have a good chance of influencing the conversation at scale.

Role 4: Analyst

This is the “what is actually working? What is actually needed?” role. We’re looking for someone diligent, thorough, and quantitative to track the effectiveness of our comms efforts as a whole.

A sampling of the sorts of things someone filling this role would do:

• Be present on social media (or regularly in contact with the people in the other roles who are monitoring social media) to keep a living list of what people are talking about with regards to AI. Bring that list back to the team on a regular basis to see what confusions the general public (or other important audiences) have, and lead the process of crafting responses.

• Help us with the process of refining memes. Run message testing, or collaborate with orgs like Palisade that are running message testing.

• Follow up with various audiences. Check comments and responses to media appearances by Nate, Eliezer, and Malo. See where people are getting lost, or where the metaphors aren’t working. Analyze MIRI’s output for stickiness, virality, and lossiness.

• Help us create concrete, actionable metrics and targets. Be the motivating force for making things grounded and systematic rather than vibes-and-intuition based. (Vibes and intuition have gotten us a long way, so far! But we can in fact do better.)

• Help us determine which audiences actually matter, in a given moment—if MIRI is spending twice as many resources as it should on Audience A, and letting Audience B lie fallow, it’s the media analyst who notices, and raises a flag.

Role 5: Pipeline builder

Much of 2025 was about refining MIRI’s memetic payloads. Nate and Eliezer created a mountain of content, and the team helped distill a lot of it into talking points and supplementals and various posts and essays.

We intend to continue creating content—there are plenty of things that have not yet been written down, or could be made clearer and more accessible. But we are currently more bottlenecked on ways to deliver, and on finding the people we need to reach, than on what, specifically, to say to them.

This role might in practice end up being the same as Role 3. But it’s possible that we should have one person building the pipelines and another person using them.

Someone serving in this role will have succeeded if, in 2026, MIRI doesn’t have to reinvent the wheel every time we ask ourselves “how do we get Message X in front of Demographic Y?” but instead have a preexisting set of “buttons” we can press. Whether our delivery methods are in-house or outsourced to allies and contractors, we need a better map of a system that is itself also better than what we have today. There are isolated cases in which someone at MIRI happens to know exactly who to reach out to if e.g. we want to put a particular thought into a popular op-ed, but there isn’t presently a clear and legible system.

Role 6: Staff Writer

As alluded to above: there are still many things left to be written. MIRI expects to be able to use more competent, motivated, high-throughput writers—people who can reliably receive writing tasks and finish them quickly, to a high standard. (This is related to but somewhat distinct from the skill of coming up with one’s own ideas as to what to write.)

Someone in this role is likely to frequently work with other teams at MIRI, such as the technical governance team, taking e.g. a draft of a technical report or proposed legislation, and creating digestible one-pagers and explainers for other audiences, or writing “press releases” that make the work comprehensible to the general public. (“Press releases” in quotes because sometimes that means a literal press release, and sometimes it means a tweet optimized for a certain kind of engagement, in conjunction with the Social Media Manager.)

We expect to have room for 1-2 staff writers who require lots of management and supervision and direction; we would likely hire more than that if we found candidates who were shaped such that they Do The Thing by default, without needing to be shepherded.

Role 7: Managing Editor

This role is more speculative, and conditional on finding multiple candidates to fill Role 6. However, if we do find ourselves with 3+ additional writers on staff, we’re highly interested in someone with experience and expertise managing a writing team—someone who is good at getting multiple people working toward a cohesive stylistic vision, and providing their writers with leads and prompts and successfully motivating them to hit their deadlines.

Recap and synthesis

In 2025, MIRI’s comms team was organized around the singular goal of making the launch of IABIED go well.

In 2026, we are looking to form a team that is taking an ambitious, multistrategy approach to getting the MIRI worldview in front of as many eyes as possible, with some degree of prioritizing particular audiences but a healthy dose of “everybody.”

Picture a team meeting in which:

• One person affirms that the newsletter is ready to go out next week, and includes the latest update from the technical governance team (who they met with yesterday). They also are tracking a conversation on the MIRI twitter account that has multiple prominent AI developers chiming in about a technical question.

• One person updates the team on the state of three separate projects: a CGPGrey collaboration, a previously-successful show writer who is working on a Netflix pitch with a script involving AI, and a funding pitch for a young up-and-comer who thinks they can turn a few thousand dollars into a series of TikToks that address two of the top ten Key Confusions on MIRI’s list.

• Speaking of those Key Confusions, another person spends a few minutes giving a report on a collaboration with Palisade in which we tried out half a dozen explanatory metaphors for Confusion Three, and we now have the one that seems to click with people best. That person will pass the new metaphor on to Nate and Malo, and in the meantime, perhaps somebody wants the MIRI account to tweet it out?

• Shifting gears, the staff writers give their updates: one of them spent the past week helping the technical governance team streamline and polish a document, and got it down to 66% as long while also making it at least 20% better. Another has been focused more on our in-group, and has a mini-sequence of three short essays ready for release on LessWrong.

This is just a single example in a pretty high-dimensional space. But what we’re getting at qualitatively is a team of largely self-directed agents, each with one or two specializations and mostly pursuing very different tasks, but all under the general umbrella of “move fast and make things (that help the overall mission).”