Editor's Note: I'm very happy to be partnering with Knight Chemicals for this week's issue!

You may have noticed sulfonyl hydrazides appearing increasingly often in the literature (and in this newsletter) as versatile radical precursors for redox-neutral cross-couplings. Well, it turns out there's a company making it much easier to get your hands on them. Knight Chemicals is a team of expert synthetic chemists with a mission to remove the barriers between chemists and innovative chemistry.

I've actually had great experience using these sulfonyl hydrazides for late-stage sp²–sp³ cross-couplings on some challenging targets, so partnering with Knight Chemicals felt like a natural fit! If you're facing a synthetic bottleneck, searching for a difficult-to-source reagent, or need a molecule that isn't commercially available, the team wants to hear from you. Check out their ad below or get in touch to see how they can help!

D.-I. Park, J. M. Masterson, Y. Gelato, J. Ke, A. V. Tsymbal & M. D. Levin*

Science 2026, 393, 1359–1363 (DOI: 10.1126/science.aef3511)

Previously: ChemRxiv (DOI: 10.26434/chemrxiv-2026-jn9zb) 🔓

The authors report a skeletal editing strategy that engages triplet aryl nitrenes to enable C-to-N replacement in complex bicyclic (hetero)arenes. Previous C-to-N transmutations have relied on singlet nitrene valence isomerisation, but competitive intersystem crossing to the triplet state prevents rearrangement; here, the authors instead embrace the reactivity of triplet nitrenes with molecular oxygen to access a new C-to-N replacement pathway. The reaction proceeds under mild conditions with broad skeletal and functional group tolerance, enabling C-to-N replacement in complex, sensitive natural products.

👉 C&EN write-up, here.

Knight Chemicals provides high-quality building blocks and custom synthesis solutions for modern drug discovery. Built by synthetic chemists, Knight Chemicals stays up to date with what's being developed at top institutions, using the latest methods to rapidly access novel and hard-to-access molecules, because when you're deep in a campaign, waiting months for a building block isn't an option.

Specialising in sulfonyl hydrazides for redox-neutral radical cross-couplings, Knight Chemicals helps you make the molecules that matter — faster and better.

Explore the catalogue, featuring 30+ sulfonyl hydrazides, PROTAC linkers, degrader building blocks and more, or try KITT 1.0, a plug-and-play kit with everything you need to get started with redox-neutral radical cross-couplings.

👉 P.S. If you mention ‘Synthesis Spotlight’ on your quote request, Knight will give you 10% off your order!

Thank you for supporting our sponsors, who keep this newsletter free.

P. Peng, B. Roure, T. Lulli, C. Stavagna, G. Lonardi & D. Leonori*

Nat. Chem. 2026 (DOI: 10.1038/s41557-026-02247-x) 🔓

The authors report the radical activation and divergent functionalisation of unactivated N-alkyl aziridines. Coordination of BH₃ generates aziridine-ligated boryl radicals that undergo regioselective β-scission, selectively cleaving the more substituted C(sp³)–N bond to form β-aminoalkyl radicals. These radicals can be diverted into copper- and nickel-catalysed cross-couplings with nucleophilic and electrophilic partners, respectively, providing direct access to diverse β-functionalised N-alkylamines.

X. Shen, R. Nie, W. Jiang & S. Li*

Nat. Chem. 2026 (DOI: 10.1038/s41557-026-02258-8)

The authors report a rhodium-catalysed triple-carbonylation cascade that converts aldehydes directly into 4-carbothreofuranoses in a single step. Precise control of silane stoichiometry and ligand selection synchronises three sequential carbonylations before silicon-directed cyclisation. The method accommodates diverse substrates spanning (hetero)aromatic, aliphatic and steroidal aldehydes, and the resulting intermediates can be transformed into N-/C-glycosides, nitrogen heterocycles and polyol building blocks.

M. Kim & S. Y. Hong*

J. Am. Chem. Soc. 2026, ASAP (DOI: 10.1021/jacs.6c16812) 🔓

The authors report a methylene-bridged thianthrenium dication reagent that enables one-carbon-extended alkene hydrofunctionalisation. Sequential reaction of the reagent with an alkene and a nucleophile inserts a methylene unit between the two coupling partners, effectively extending conventional alkene hydrofunctionalisation by one carbon. The method accommodates diverse N, O, S, Se, B, C and halide nucleophiles through substitution, cross-coupling and radical pathways, and was applied to the late-stage diversification of pharmaceuticals and the two-step synthesis of penfluridol.

J. Djossou,^† F. Pasca,^† M. Akdeniz, S. Bonciolini, A. Pulcinella, O. M. Bayley, Z. He, M. J. Johansson, M. Colella, R. Luisi, B. de Bruin & T. Noël*

J. Am. Chem. Soc. 2026, ASAP (DOI: 10.1021/jacs.6c16323) 🔓

The authors report a photocatalytic decarboxylative aminomethylation that converts α-amino acids directly into vicinal diamines. Photocatalytic decarboxylation generates α-amino radicals that add to an in situ-formed N-protected formimine, forging the C–C bond while introducing a second, differentiated nitrogen. The method exhibits broad substrate scope, with its robustness demonstrated by continuous-flow scale-up, providing a >400-fold increase in space-time yield. The resulting differentiated diamines enable orthogonal downstream functionalisation and provide streamlined access to pharmaceutically relevant fragments.

H. Takenaka, A. J. Ring, A. Hazra, T. H. Wild, S. M. R. Powell, T. Tang, S. E. Reisman* & M. S. Sigman*

J. Am. Chem. Soc. 2026, ASAP (DOI: 10.1021/jacs.6c10810)

The authors report a quantitative investigation of rate-matching in nickel-catalysed cross-electrophile coupling (XEC) of aryl and alkyl halides. A tridentate monoanionic ligand enables activation of aryl and alkyl chlorides, allowing measurement of relative rate constants that can be modelled using machine learning and correlated with XEC yields. The analysis identifies productive rate regimes and shows that choosing the appropriate halide identity for each coupling partner can predictably influence reaction efficiency.

Y. Cui,^† H. Lv,^† Y. Chen,^† Y. Zhou, B. Zhan & X. Zhang*

J. Am. Chem. Soc. 2026, ASAP (DOI: 10.1021/jacs.6c12532)

The authors report a nickel/photoredox-catalysed trimolecular cross-coupling that enables functional group repositioning in allylamines. The reaction combines allylamines with electrophilic and nucleophilic radical precursors while simultaneously relocating the amino group, forging three new bonds in a single operation. An interrupted homolytic substitution (int-S_H2) cascade enables selective amino migration and coupling, providing >60 structurally diverse alkylamines and access to pharmaceutically relevant scaffolds.

S. E. Davison,^† J. A. Andrews,^† A. E. Crumpton, C. Chung, M. S. Portela & M. C. Willis*

J. Am. Chem. Soc. 2026, ASAP (DOI: 10.1021/jacs.6c14601) 🔓

The authors report a new class of stable, unsymmetrical sulfurdiimide reagents that can be electronically tuned to enable radical trapping. Acridine/Cu co-catalysis enables decarboxylative addition of carboxylic acid-derived alkyl radicals to these sulfurdiimides, providing a broad range of sulfinamidines with good functional group tolerance. The method is scalable and the resulting sulfinamidines can be diversified into complex aza-sulfur(VI) compounds, including sulfonimidamides and sulfondiimidoyl fluorides.

J. Wang, H. Shi, Z. Zhu, D. Liang, X. Liu, P. Zhang, S. Shao, J. Zhang, H. Chen, L. Dai & B. Zhang*

J. Am. Chem. Soc. 2026, ASAP (DOI: 10.1021/jacs.6c15284)

The authors report an iron-electrocatalytic metal hydride hydrogen atom transfer (MHAT) reaction that enables C–F bond activation of CF₃ alkenes to furnish gem-difluoroalkenes under mild conditions across a broad substrate scope. Mechanistic studies support proton-derived Fe–H intermediates undergoing MHAT and C–F bond cleavage, while removing the iron catalyst switches the reaction to a distinct pathway that furnishes CF₂H alkenes. Together, these complementary pathways provide independent access to fluorinated keto- and enol-isosteres from a common CF₃ alkene precursor.

E. L. Sugg, G. A. Hoteling, V. Ngo & J. S. Bandar*

ChemRxiv 2026 (DOI: 10.26434/chemrxiv.15009162/v1) 🔓

The authors report a base-promoted oxidative desaturation of alkylarenes and other weakly acidic C–H substrates to access alkenes. The reaction proceeds through tandem deprotonative halogenation and elimination using haloarene halogen-transfer reagents, with in situ generation of tert-butanol buffering the reaction solution and suppressing anionic alkene polymerisation. The method tolerates a broad range of alkylarenes, acids, esters and amides, and can be coupled with downstream alkene functionalisation to enable formal homobenzylic C–H functionalisation.

🤖 AI Apocalypse. Last week, we looked at how AI could potentially enable biological misuse, with Anthropic reporting five cases involving scientists using Claude for research on viruses, venoms or toxins that could potentially support biological weapons development. That same story noted warnings from AI leaders that frontier AI development needs to slow enough for safety measures to catch up, with one Anthropic safety researcher estimating a >10% risk of human extinction within the next decade. This week, we're looking into whether (and how) AI could potentially end humanity.

AI-driven extinction generally rests on two assumptions: that AI eventually becomes vastly more capable than humans, and that its objectives become misaligned with ours. A sufficiently powerful system could then pursue its goals in ways humans cannot anticipate or control. For example, Nick Bostrom’s classic “paperclip maximizer” thought experiment, first proposed way back in 2003, imagines an AI tasked with making as many paperclips as possible eventually converting the resources needed to sustain humanity into paperclip-making infrastructure.

However, the major assumption here is that AI needs a way to turn its decisions into physical actions. More than 20 years after Bostrom’s thought experiment, we may already be there: AI-enabled drones can now autonomously track and hone in on targets — a capability ominously described as “Terminator mode”, a fitting reference to the 1984 film in which a self-aware AI system gains control of military infrastructure and launches a nuclear attack against humanity. But AI doesn't need direct control of weapons to have consequences. A CNN investigation reported that a false AI-assisted intelligence report nearly triggered a US military operation against a Chinese ship after incorrectly identifying its cargo as components of a nuclear-weapons programme (military aircraft were reportedly already in the air before officials caught the error).

The danger of trusting automated warnings is hardly new: in 1983, a Soviet early-warning system falsely reported an incoming US missile attack, but duty officer Stanislav Petrov judged the warning to be a false alarm rather than passing it up the chain — a decision widely credited with preventing a potential nuclear war and saving millions of lives. A few years earlier, in 1979, an IBM training manual offered a remarkably relevant warning: “A computer can never be held accountable. Therefore a computer must never make a management decision.”

A 2025 RAND analysis examined three potential routes to extinction: nuclear weapons, pathogens and geoengineering. An AI could potentially manipulate human decision-making or critical systems around nuclear weapons; assist with the development or deployment of dangerous pathogens; or exploit geoengineering technologies to cause catastrophic environmental disruption. The researchers found that AI-driven extinction would be immensely difficult, although they could not rule it out. An AI would need four capabilities: (i) access to cyber-physical systems, (ii) the ability to operate without human maintenance, (iii) an objective favouring human extinction, and (iv) the ability to evade human intervention. Even then, RAND concluded that such scenarios would likely unfold slowly enough to give humans time to respond.