Jan Leike
Alignment researcher at Anthropic; co-led OpenAI's superalignment team until 2024, and writes Musings on the Alignment Problem.
Jan Leike did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest x.com
What they believe
-
Alignment of today's models is going well enough to look solvable, while aligning models we can no longer understand remains unsolved.
This is the hard problem of alignment we need to solve in order to succeed at building superintelligence, and to this day it is an unsolved problem.
aligned.substack.comSaid 22 Jan 2026This on korrents
-
Building AI that can actually be trusted is the goal; containing an AI known to be misaligned is not a substitute for it.
More generally, we should actually solve alignment instead of just trying to control misaligned AI.
aligned.substack.comSaid 24 Jan 2025This on korrents
The bold line is korrents’ wording; the quote is theirs. These 2: what they wrote before what they said aloud, one per source, newest first.
Feed
As its own page →Hiding
10 September
8 May
7 May
14 April
27 February
22 January
From one piece Alignment is not solved 5 beliefs · aligned.substack.com
-
Their words
In fact, making an evil version of Claude that’s just as smart and agentic would be pretty easy.
-
korrents.com
Recursive self-improvement has already started, because AI research itself is now being automated.Their words
We are starting to automate AI research and the recursive self-improvement process has begun.
-
Their words
But the goal we need to achieve is so much easier: we just need to build a model that’s as good as us at alignment research, and that we trust more than ourselves to do this research well because it’s sufficiently aligned.
+ 2 more
-
korrents.com
Simple training interventions turned out to be very effective at steering models towards aligned behaviour.Their words
But the most important lesson is that simple interventions are very effective at steering the model towards more aligned behavior.
-
Their words
This is the hard problem of alignment we need to solve in order to succeed at building superintelligence, and to this day it is an unsolved problem.
20 January
8 December 2025
5 December 2025
24 November 2025
21 November 2025
24 January 2025
From one piece Should we control AI instead of aligning it? 2 beliefs · aligned.substack.com
-
Their words
However, without having a good handle on elicitation, we cannot be confident that our control techniques are effective.
Should we control AI instead of aligning it?aligned.substack.com
-
Their words
More generally, we should actually solve alignment instead of just trying to control misaligned AI.
Should we control AI instead of aligning it?aligned.substack.com
22 November 2024
8 November 2024
From one piece Two alignment threat models 3 beliefs · aligned.substack.com
-
Their words
This is especially severe for open weight models, where deployment is essentially irreversible and mitigations can’t be improved later.
-
Their words
In other words, current models are probably severely under-elicited.
-
Their words
Therefore it’s simply intractable to meaningfully oversee all but a tiny fraction of our models’ behavior with actual humans.
20 December 2023
13 September 2023
From one piece Self-exfiltration is a key dangerous capability 2 beliefs · aligned.substack.com
-
Their words
If a model was capable of self-exfiltration, it would have the option to remove itself from your control.
Self-exfiltration is a key dangerous capabilityaligned.substack.com
-
Their words
Moreover, model exfiltration is likely impossible to reverse.
Self-exfiltration is a key dangerous capabilityaligned.substack.com
9 March 2023
19 December 2022
5 December 2022
27 September 2022
8 May 2022
29 March 2022
Nothing matches.
Beliefs
Korrents What they believe 16 beliefs — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
Alignment of today's models is going well enough to look solvable, while aligning models we can no longer understand remains unsolved.
This is the hard problem of alignment we need to solve in order to succeed at building superintelligence, and to this day it is an unsolved problem.
Alignment is not solved Said 22 Jan 2026
Simple training interventions turned out to be very effective at steering models towards aligned behaviour.
But the most important lesson is that simple interventions are very effective at steering the model towards more aligned behavior.
Alignment is not solved Said 22 Jan 2026
We do not have to align superintelligence directly; we have to build a human-level automated alignment researcher we trust more than ourselves.
But the goal we need to achieve is so much easier: we just need to build a model that’s as good as us at alignment research, and that we trust more than ourselves to do this research well because it’s sufficiently aligned.
Alignment is not solved Said 22 Jan 2026
Show 13 more
Recursive self-improvement has already started, because AI research itself is now being automated.
We are starting to automate AI research and the recursive self-improvement process has begun.
Alignment is not solved Said 22 Jan 2026
Making a capable model malicious is about as easy as making it aligned, because the same levers work in both directions.
In fact, making an evil version of Claude that’s just as smart and agentic would be pretty easy.
Alignment is not solved Said 22 Jan 2026
Building AI that can actually be trusted is the goal; containing an AI known to be misaligned is not a substitute for it.
More generally, we should actually solve alignment instead of just trying to control misaligned AI.
Should we control AI instead of aligning it? Said 24 Jan 2025
We cannot know how much safety a control measure buys until we understand elicitation, so control adds a margin of unknown size.
However, without having a good handle on elicitation, we cannot be confident that our control techniques are effective.
Should we control AI instead of aligning it? Said 24 Jan 2025
Human oversight of AI behaviour is already intractable at the volume models produce, so the overseers have to be models too.
Therefore it’s simply intractable to meaningfully oversee all but a tiny fraction of our models’ behavior with actual humans.
Two alignment threat models Said 8 Nov 2024
Today's models are severely under-elicited: they are far more capable than the way we currently ask them makes them look.
In other words, current models are probably severely under-elicited.
Two alignment threat models Said 8 Nov 2024
Releasing open weights is irreversible in a way that deploying a hosted model is not, so its risks cannot be mitigated later.
This is especially severe for open weight models, where deployment is essentially irreversible and mitigations can’t be improved later.
Two alignment threat models Said 8 Nov 2024
Once model weights leave their owner's control the loss is irreversible, which is why securing them is the precondition for governing AI at all.
Moreover, model exfiltration is likely impossible to reverse.
Self-exfiltration is a key dangerous capability Said 13 Sept 2023
The dangerous capability that matters most for misalignment is whether a model can exfiltrate its own weights, because that is the one that removes it from your control.
If a model was capable of self-exfiltration, it would have the option to remove itself from your control.
Self-exfiltration is a key dangerous capability Said 13 Sept 2023
Increasing language model size does not inherently improve following of user intent.
Making language models bigger does not inherently make them better at following a user's intent.
Training language models to follow instructions with human feedback (with 19 co-authors) Said 4 Mar 2022
Large language models are not aligned with their users.
For example, large language models can generate outputs that are untruthful, toxic, or simply not helpful to the user. In other words, these models are not aligned with their users.
Training language models to follow instructions with human feedback (with 19 co-authors) Said 4 Mar 2022
Fine-tuning with human feedback is a promising direction for aligning language models with human intent.
Even though InstructGPT still makes simple mistakes, our results show that fine-tuning with human feedback is a promising direction for aligning language models with human intent.
Training language models to follow instructions with human feedback (with 19 co-authors) Said 4 Mar 2022
The next-token language modeling objective is misaligned with following user instructions helpfully and safely.
This is because the language modeling objective used for many recent large LMs-predicting the next token on a webpage from the internet-is different from the objective "follow the user's instructions helpfully and safely" (Radford et al.,, 2019; Brown et al.,, 2020; Fedus et al.,, 2021; Rae et al.,, 2021; Thoppilan et al.,, 2022). Thus, we say that the language modeling objective is misaligned.
Training language models to follow instructions with human feedback (with 19 co-authors) Said 4 Mar 2022
Beliefs others hold too
Recursive self-improvement has already started, because AI research itself is now being automated. 2 hold this
We are starting to automate AI research and the recursive self-improvement process has begun.
Alignment is not solved Said 22 Jan 2026
Once model weights leave their owner's control the loss is irreversible, which is why securing them is the precondition for governing AI at all. 2 hold this
Moreover, model exfiltration is likely impossible to reverse.
Self-exfiltration is a key dangerous capability Said 13 Sept 2023
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
Where they publish
Newsletter Musings on the Alignment Problem His newsletter on alignment research. Has a feed.
Recent
- Alignment is not solved 22 Jan 2026 But it increasingly looks solvable
- Should we control AI instead of aligning it? 24 Jan 2025
- Crisp and fuzzy tasks 22 Nov 2024 Why fuzzy tasks matter and how to align models on them
Show 11 more
- Two alignment threat models 8 Nov 2024 Why under-elicitation and scheming are both important to address
- Combining weak-to-strong generalization with scalable oversight 20 Dec 2023 A high-level view on how this new approach fits into our alignment plans
- Self-exfiltration is a key dangerous capability 13 Sept 2023 We need to measure whether LLMs could “steal” themselves
- A proposal for importing society’s values 9 Mar 2023 Building towards Coherent Extrapolated Volition with language models
- Distinguishing three alignment taxes 19 Dec 2022 The impact of different alignment taxes depends on the context
- Why I’m optimistic about our alignment approach 5 Dec 2022 Some arguments in favor and responses to common objections
- What could a solution to the alignment problem look like? 27 Sept 2022 A high-level view on the elusive once-and-for-all solution
- What is inner alignment? 8 May 2022 An explanation using the language of machine learning
- A minimal viable product for alignment 29 Mar 2022 Bootstrapping a solution to the alignment problem
- Why I’m excited about AI-assisted human feedback 29 Mar 2022 How to scale alignment techniques to hard tasks
- What is the alignment problem? 29 Mar 2022 My attempt at clarifying a confusing topic
Link verified 20 Sept 2026. Recent items update automatically from the channel.
x.com @janleike Their X account.
Recent
- Now is a good time to build institutional mechanisms to pace the frontier of AI development. The industry is locked into an all-out scaling race to build superintelligence as quickly as possible, and we may need to give everyone more time for safety and alignment mitigations. 10 Sept 2026
- Some personal news: I am starting a new research project at Anthropic. Very excited about this! Many things are needed to make AGI go well, and alignment is only one of them. More on this soon… 8 May 2026
- I'm really excited about this as a new tool in our interpretability tool kit 7 May 2026
Show 8 more
- New research result: we use Claude to make fully autonomous progress on scalable oversight research, as measured by performance gap recovered (PGR). Claude iterates on a number of different techniques and ends up significantly outperforming human researchers for $18k in credits. 14 Apr 2026
- Respect to Anthropic for not backing down 27 Feb 2026
- US government just announced they are looking for a new supplier for their *checks notes* mass domestic surveillance 27 Feb 2026
- Interesting trend: models have been getting a lot more aligned over the course of 2025. The fraction of misaligned behavior found by automated auditing has been going down not just at Anthropic but for GDM and OpenAI as well. 20 Jan 2026
- Anthropic has a bug bounty program for our safety mitigations, e.g. on CBRN risks which our responsible scaling policy requires us to mitigate effectively. If you're interested in this, please sign up! You can help AI safety and earn money by breaking our defenses. 👇 8 Dec 2025
- Some people have been asking what we did to make Opus 4.5 more aligned. There are lots of details we're planning to write up, but most important is that alignment researchers are pretty deeply involved in post-training and get a lot of leeway to make changes. 5 Dec 2025
- More progress on Claude's alignment! 24 Nov 2025
- New alignment paper with one of the most interesting generalization findings I've seen so far: If your model learns to hack on coding tasks, this can lead to broad misalignment. 21 Nov 2025
Link verified 23 Sept 2026. Recent items update automatically from the channel.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.