The Round-Up #2
Dire wolves, AI virologists, sci-fi TV, and much more!
Welcome (or welcome back) to The Science Fictional Now, where we discuss biotech, the history of science, and science fiction - sometimes all at once! I’m excited to be back, and with a post chock full of interesting happenings from the past several months. In the first half of this Round-Up, I’ll break down a some of these in a few paragraphs each; this includes the controversial dire wolf resurrection episode that took the internet by storm a few months back. The second half is a classic Substack link post, replete with essays, interviews, and news stories worth checking out. Feel free to skip around using the table of contents below, or read the post straight through.
Table of Contents
The return of the dire wolf…?
“This is Remus. He’s a dire wolf. The exist in over 10,000 years.” So read the cover of Time magazine just over two months ago, when Dallas-based Colossal Biosciences announced that they had resurrected the long-extinct canine. But had they? Best known as the House of Stark’s sigil in Game of Thrones, the dire wolf was a real animal that roamed the Americas for millennia before going extinct around 10,000 years ago. The story’s pop culture relevance and scientific novelty earned it a flurry of media coverage, but Colossal’s claims also came under scrutiny from biologists. Now that the dust has settled a little bit, I’ll break down exactly what happened - including both the science and the controversy - and offer a bit of commentary on the whole affair.
The science
Understanding why Colossal’s dire wolf claims ended up being controversial requires and understanding of the science behind the canines’ births. In the Time exclusive article, Colossal CEO Ben Lamm revealed that the company taken cells from the bodies modern grey wolves and then edited them using CRISPR to more closely resemble dire wolf genomes that the company had previously reconstructed from fossils. Once the cells were edited, they could be used to start pregnancies through a process called somatic cell nuclear transfer cloning - the same procedure famously used to create Dolly the Sheep in 1996. Colossal scientists extracted unfertilized eggs from grey wolves, replaced their nuclei with those of the edited cells, and re-inserted them into surrogate mothers. Ultimately, three pups were born: brothers Romulus and Remus and little sister Khaleesi (named for Game of Thrones’ Daenerys Targaryen).
Since the dire wolf is extinct though, Colossal scientists had to reconstruct its genome before they could start editing anything. This meant carefully collecting samples from dire wolf fossils and extracting the DNA, which they could then sequence and compare to the genetic code of modern wolves. None of this ancient DNA was inserted into cells though; Colossal didn’t so much resurrect dire wolves as recreate them. I think that this is an important distinction to make, and I’ll refer to Colossal’s wolves as “neo-dire wolves” accordingly.
Colossal’s editing strategy was to pinpoint key differences between the grey wolf and dire wolf genomes in order to recreate dire wolves’ most distinctive traits with a small number of edits. Ultimately, they made 20 changes affecting 14 genes using a CRISPR-based technology called base editing, which alters to individual DNA bases without breaking the double helix apart.1 Intriguingly though, none of these edits fell within the genes themselves. Instead, Colossal exclusively exclusively edited gene enhancers, stretches of DNA that influence how much each gene is read inside cells. The neo-dire wolves actually have all the same proteins as modern grey wolves, just in different amounts. To me, this is both an intriguing and impressive result.
The fallout
Of course, making just 20 changes meant that Colossal’s neo-dire wolves still had a massive number of differences from the dire wolf genomes reconstructed by the company. These genomes had around 99.5% similarity to modern grey wolf DNA, but this still meant that there were more than 12 million differences. A number of scientists swiftly disputed Colossal’s claim that they had resurrected an entire species.
But what exactly is a species anyway? The definition that you’ll find in most textbooks stipulates that species are groups of organisms that can interbreed with each other; this “biological species concept” was first defined by evolutionary biologist Ernst Mayr in 1942.2 It’s unlikely that Colossals’ neo-dire wolves meet this bar, since they’d need to have been edited in a way that disabled mating with normal grey wolves. Mayr’s definition of species isn’t perfect though, and breaks down when applied to phenomena such as ring species and asexual reproduction.
In fact, if anything is true about definitions of species, it’s that there are lots of them. Creating a definition of species that works for all organisms is actually quite difficult and Mayr’s famous solution is just one part of a centuries-old conversation between evolutionary biologists, paleontologists, and geneticists. Colossal has used the “morphological” definition of species to argue that their neo-dire wolves are distinct from modern grey wolves. The basic idea is that if Romulus, Remus, and Khaleesi look and act like the dire wolves of old, they should be recognized that way regardless of their genetic content or origin. It’s worth noting that the wolves are still adolescents, so it’s technically too early to say whether they’ll recapitulate certain key traits, such as dire wolves’ notable size.
Moreover, it’s a bit ironic that Colossal continues to rely on the morphological definition given that their entire approach was enabled by cutting edge genome sequencing and editing tools. Accessible DNA sequencing has made it easier to classify species and determine evolutionary relationships by comparing their genomes, making this approach logical for a company like Colossal. But given the small number of edits they made, it’s very likely that their neo-dire wolves would be much closer to grey wolves than the dire wolves of old on a phylogenetic tree. This was the main reason so many scientists insisted Colossal’s canines shouldn’t be called dire wolves.
In the wake of the controversy, Colossal CSO Beth Shapiro claimed that the company had always been honest about their wolf pups’ origins, telling New Scientist that “Our animals are grey wolves with 20 edits that are cloned…And we’ve said that from the very beginning. Colloquially, they’re calling them dire wolves and that makes people angry.” Again, Colossal’s story broke in an exclusive with Time, so they had ample opportunity to dictate the terms of discourse. While Colossal’s PR team certainly hasn’t backed off, I’m inclined to agree with those who say Shapiro has changed her tune a bit - perhaps an indication that the dire wolf detractors have mostly prevailed in academic circles. More than anything, I think the whole affair has revealed one way in which advances in genetics and genomics will continue to influence the conversation on how species ought to be defined.
The real goal
At the end of the day though, Colossal’s dire wolf project was first and foremost a publicity stunt - and quite a successful one at that. Colossal’s stated goal is to develop genomics tools for conservation purposes, protecting endangered animals and recreating extinct ones like the wooly mammoth. The earth looked quite different when dire wolves went extinct 10,000 years ago though, and the ecological benefits of brining back such and ancient species are unclear. Shapiro addressed how the dire wolf project relates to Colossal’s broader vision for an MIT Technology Review article:
“The motivation really is to develop tools that we can use to stop species from becoming extinct. Do we need ancient DNA for that? Maybe not,” says Shapiro. “Does it bring more attention to it so that maybe people get excited about the idea that we can use biotechnology for conservation? Probably.”
So far, it looks like the main conservation benefit of the project was actually the cloning of the critically endangered red wolf, which most media outlets mentioned in passing if at all. This type of lower-profile impactful work doesn’t seem to be Colossal’s focus though; if anything it’s to inspire good will and enable the flashy moonshot de-extinction projects. Some ecologists have even argued that the work is counterproductive because it distracts from more pressing conservation issues, such as the real-time loss of biodiversity. In fact, the project might have increased extinction risk for other species already: Interior Secretary Doug Burgum cited the work as a reason to relax efforts to preserve already endangered species.3
Colossal has also filed a number of dire wolf-related trademarks, including merchandising rights for the cubs’ likenesses, and applied for a patent on the species itself. While Lamm stated that the company has no current plans to sell dire wolf apparel, he also recently hinted at a collaboration between his company and the upcoming film Jurassic World: Rebirth. This isn’t totally surprising given how much Colossal benefitted from the Game of Thrones association, though a Jurassic Park connection might be a bit on the nose.
But what’s clear either way is that Colossal thinks they can continue to succeed through spectacle. It’s no coincidence that species are being “resurrected” in a world where billionaires like Bryan Johnson fund research longevity research to advance their personal vanity projects. Both Johnson’s personalized longevity regime and Colossal’s dire wolf project are set up to benefit them first, with the vague implication that success will trickle down to others later. I’m not against longevity research and I’d be open to de-extinction projects with clearly articulated benefits - I just don’t like the idea of making money first and worrying about bioethics never.
Colossal has done some interesting science and created some cute wolf cubs, but I’m far from convinced that their approach represents the best path forward for conservation biotech. I really do hope that their experience making dire wolves helps safeguard endangered species one day - I’m just not holding my breath.
Thanks to Kevin Wasko for relevant discussions.
AI virologists
From videos on social media to the corporate world and even discussions of a coming singularity, it’s hard to deny that AI has generated considerable buzz. A slightly less visible area where machine intelligence has become of considerable interest is biosecurity - the field concerned with preventing pandemics and bioterrorism. A particular worry in the community is that large language models (LLMs) could help would-be bioterrorists gain knowledge necessary to develop enhanced pathogens, including viruses. To assess the current level of risk, researchers at SecureBio - a Boston-based nonprofit research organization - recently developed a test to benchmark LLMs’ virology knowledge.
SecureBio’s benchmarking assessment is called the Virology Capabilities Test (VCT) and consists of over 300 multiple choice questions related to virology. When SecureBio researchers asked the questions to contemporary LLMs, they found that a number of the models outperformed virology experts (mostly grad students and postdocs), even on questions related to the experts’ specific areas of focus. This is a bit of a double-edged sword. While it indicates that AI might help advance virology research that prepares us for future pandemics, it also lends some credence to worries that LLMs might dispense sensitive information.
AI news has a tendency to attract polarized responses, and the VCT unsurprisingly elicited both concerned responses and heavy skepticism. However, I think the story deserves more moderate reactions, especially considering that the VCT consists of multiple choice questions. While the LLMs performed well on the test, this didn’t require them to troubleshoot virology experiments, generate protocols, or articulate tacit scientific knowledge - all things that would provide much more uplift to untrained bioterrorists.
This isn’t necessarily an issue, though. By using multiple choice questions, SecureBio team picked a tractable way to assess LLMs’ virology knowledge and put it into action. Their preprint helps us understand the problem better and underscores the need for further work; the paper is informative but shouldn’t be the end of the story. In my view, this was perfectly reasonable research that got blown out of proportion by knee-jerk responses.
LLMs’ biology knowledge will only deepen in the future, so benchmarks like the VCT can make a huge difference if they’re interpreted correctly and deployed early. Unfortunately, the past several months have been a bit of a mixed bag for AI biosecurity advocates. On one hand, multiple AI companies have implemented new biorisk safeguards and researchers have published roadmaps for enhancing security even further. Conversely, some companies’ technical updates have drawn public criticism and a number of key players in AI have staunchly opposed recent attempts to regulate of the industry. Technical AI safety research like the VCT isn’t sufficient to ensure safe AI development but, if deployed correctly, could could be a crucial tool in pushing companies and policymakers in the right direction.
Screen sci-fi is heating up
Search the internet for takes on the current state of Hollywood and it probably won’t take you long to encounter the word “slop.” The world neatly encapsulates an increasingly common attitude towards movies and TV today: the dominance of “cinematic universes” and franchises, generic Netflix flicks, and generally dopamine-fueled media has created a landscape replete with shallow, low-effort entertainment. I’m not here to dispute this trend. Rather, my goal is to highlight one area of entertainment that I see as a diamond in the rough: science fiction TV.
I’ll freely admit that I’m a bit biased here. It’s no secret that I love science fiction and on the screen that includes everything from Star Wars to Solaris. This is due in no small part to the diversity of the genre; the freedom to imagine distant futures and new worlds entirely gives science fiction narrative and thematic flexibility as well as limitless stylistic and aesthetic possibilities. This diversity has long been evident in science fiction literature - a point I highlighted in a previous post on entry points to the genre - but screen sci-fi has mostly been a subgenre of action-adventure.4 Fortunately, I think that’s begun to change, and the first half of 2025 has provided some wonderfully creative science fiction TV.
For starters, Black Mirror, which released its seventh series in April, has made its particular brand of near-future dystopia into a highly recognizable form of science fiction. In my experience, it’s not uncommon to hear people say that something “feels like Black Mirror.” With the release of its second season, Severance went from being an under-the-radar mystery box show to a legitimate pop-culture hit. Star Wars was always about fighting authoritarianism, but Andor’s second season brought this clearer into focus through its gritty atmosphere and naturalistic approach.5 Netflix’s Eternaut brought the classic Argentine comic to the screen, keeping the original Spanish while updating the setting and political commentary for the present day. Other shows that have expanded the palette of sci-fi TV in the past five years include For All Mankind, Scavenger’s Reign, and Devs.
This diversification of science fiction TV hasn’t occurred in a vacuum though, and therefore isn’t isolated from other media trends. For example, I’d say that shows like Foundation and The Peripheral have helped expand the genre, but I’d also argue that their writing hasn’t matched the strength of their casts and special effects. Other creative projects that I thought deserved more attention simply failed to find audiences: the Paper Girls TV adaptation was quickly cancelled after a single season while in the movie world, Furiosa flopped at the box office. Lastly, the majority of the releases that I’ve mentioned so far were tied to existing IPs, either as adaptations or sequels.
But why is science fiction TV in particular expanding its horizons and rising above the “slop” status quo? Maybe superhero media has optimized futuristic action movies for dopamine production so well that science fiction has actually been freed up to explore new directions. Maybe the race to adapt every popular IP has actually unintentionally led screen science fiction to better represent the genre’s diverse print output. I won’t pretend that I have a concrete answer to this question - really, this whole section is a largely qualitative think piece with lots of personal preference involved. My goal here isn’t to crunch viewership numbers but simply explain why I can’t help but feel a little bit hopeful about the future of science fiction TV. If you’re feeling disillusioned about today’s media landscape, I’d suggest giving some of the shows mentioned here a look.
Quick hits
Some salient thinking on AI
Arvind Narayanan and Sayash Kapoor discussed the disadvantages of AI over-reliance in science in a commentary piece for Nature.
Abi Olvera’s article in the Bulletin of the Atomic Scientists detailed other metascience barriers that will likely limit even AI-accelerated science.
Ted Chiang’s New Yorker essay from last summer is probably the most considered take on AI art that I’ve encountered.
Science and the government
Donald Trump’s second presidency has brought about so many radical science policy changes that it would be difficult to cover them all - I’ll focus on a few key ones here.
One huge issue is the massive curtailment of funding for basic science - Matteo Wong did an excellent job explaining why this is a problem in a recent Atlantic article. For example, commercial tech development requires continuous scientific stimulus, industry researchers are trained using federal dollars, and the government maintains databases and other public goods that benefit private companies.
Despite these funding cuts, maintaining US biotech leadership remains a goal of the Trump administration. A recent report from the National Security Commission on Emerging Biotechnology set forth some good ideas on this front; the Council on Strategic Risks also put out a nice summary of report key points.
Trump also signed an Executive Order restricting “gain of function” (GoF) research - science with the goal of improving or adding functions to an organism - on pathogens. GoF studies on dangerous agents can make them even more hazardous, but can also biology useful in developing therapeutics and diagnostics.
I’m in general a fan of stricter regulation of GoF research and I like some points of the executive order, such as requiring federally-funded GoF research to be reported to the government, and restricting funding for GoF studies in countries without proper oversight capacity.
Kai Kupferschmidt’s piece for Science illustrated the key issue with the Executive Order though, which is that it uses an overly broad definition of GoF research. Many biosecurity researchers accordingly feel that the Order actually makes the US more susceptible to biological risks.
The Order also fits within the Trump administration’s larger distrust of science, a point emphasized in a recent editorial from Global Biodefense.6 Priorities implicated in the administration’s adherence to SARS-CoV-2 lab leak theories and other initiatives are therefore likely to manifest in the GoF Order’s implementation.
Some paleogentics fun
Traditionally, Neanderthals were thought to have been “absorbed” by modern humans after first making contact, but new research is showing that there were multiple sustained interactions between the two groups. This is detailed in a recent Science paper from Joshua Akey’s lab at Princeton, and broken down in a research highlight from the university.
Paleogeneticist David Reich more broadly discussed the complicated geography of early human genetics on Dwarkesh Patel’s podcast (h/t Chase Chiang).
The first individualized CRISPR therapy: In a milestone for genome editing, researchers developed a custom treatment for a baby with carbamoyl-phosphate synthetase 1 deficiency, a rare but severe disease. Eric Topol explained this nicely on his own Substack.7
Engineering dragons: Cathy Tie and Josie Zayner recently announced the Los Angeles Project, which has the goal of creating new animals - including real-life unicorns and dragons - with genome editing. My thoughts are basically in line with Paul Knoepfler’s take on the matter.
The personal side of genetic disease: Jennifer Lai Remmel wrote about genetic disease’s effects on her family, how it motivated her own research career, and her successful IVF story in a recent piece for Asimov Press.
Revenue sharing: Jay Keasling’s lab at UC Berkeley recently devised a synthetic pathway for a therapeutic compound traditionally isolated from tree bark. The molecule, called QS-21, is the active ingredient in traditional medicines, and Keasling’s group worked out an agreement to share proceeds with the indigenous communities responsible for these treatments.
Multitudes of ideas: Science writer Ed Yong sat down for an insightful podcast interview with the New York Times, where he discussed his experience reporting on COVID-19, writing about microbes, and becoming a birding fanatic.
Prions, folding, and protein design: The popular perception of prions isn’t exactly positive; they’re mostly known as the pesky misfolded proteins that cause mad cow disease and Creutzfeldt-Jakob syndrome. However, Eryney Marrogi and Theodore Sternlieb recently argued that bioengineers should study prions’ unique folding properties in order to get better at designing enzymes with new functions. For non-scientists, the article is worth reading simply for its history of prion research, which was initially quite a controversial field.
Bridging the gap in genome editing: In my last Round-Up and a recent article for Berkeley Science Review, I discussed a new class of genome editors called bridge RNAs. Since then, researchers have demonstrated that these complexes work in human cells, paving the way for diverse therapeutic and research applications. Niko McCarty recently recapped this new development on his Substack.
More shameless self-promotion: In addition to the bridge RNA article, I also recently penned a piece about the life of physician-scientist-writer Lewis Thomas for Princeton Alumni Weekly, as well as a history of tomato breeding for Notes in Progress, and a general audience overview of CRISPR genome editing for Asimov Press.
Shortly after the initial Time story broke, Colossal posted a preprint detailing the genomics side of the dire wolf work. While this focused on dire wolf ancestry and didn’t go into the genome editing and cloning work, the company’s press release on BusinessWire discussed many of the specific gene targets. 15 of the 20 edits made were taken directly from the reconstructed dire wolf genome.
This can theoretically be caused by a single mutation so arguments claiming that dire wolves aren’t different enough from grey wolves to be their own species should be made using other definitions of species.
This isn’t to say that there weren’t dramatic, philosophical, or otherwise creative science fiction movies and TV until recently. My point is simply that sci-fi taking a different approach from Independence Day or the original Battlestar Galactica has increased both in volume and prominence.
The Last of Us is another great example of the trend I’m describing here but given the “controversy” around its recent second season I couldn’t work it in concisely enough. I don’t think season two was perfect but it’s been dragged down by a lot of bad-faith criticism, some of which was applied to its video game precursor and some of which is new. The Last of Us: Part 2 was an important and thematically necessary follow-up to the first game in my opinion, and while I thought the show’s second season was mostly a good adaptation, the decision to cut it off in the middle didn’t work for me at all.
I do think this piece is overly defense of GoF research’s current status quo.
I haven’t heard anyone else call base editing “CRISPR 2.0” and I don’t think it’s that useful of a term personally.