Can You Turn Other Elements Into Gold? Alchemy to Nuclear Transmutation

No chemical reaction can turn lead, mercury or anything else into gold, because chemistry rearranges electrons and an element is defined by the protons in its nucleus. Nuclear transmutation can make gold, and physicists have done it since 1941, but only in trace amounts, at a cost many orders of magnitude above the market price, and usually as radioactive isotopes that decay away within days.

Below: why, every documented case of man-made gold from the original papers, an unproven 2025 fusion proposal, and the gold you can actually buy, from karats and plating to hallmarks and scams. Nothing here is a method for anything hazardous or radioactive.

Key takeaways

• An element is defined by its proton count. Gold has 79 protons, mercury 80, lead 82 and bismuth 83. Changing that number is a nuclear process, not a chemical one.
• In 1941, Harvard researchers Sherr, Bainbridge and Anderson reported gold made from mercury using fast neutrons. The gold was radioactive, with half-lives of 65 and 78 hours.
• In work submitted in August 1980 and published in Physical Review C in March 1981, a Lawrence Berkeley Laboratory team including Glenn Seaborg measured gold isotopes made by smashing carbon and neon nuclei into bismuth.
• In May 2025, CERN's ALICE collaboration reported that near-miss lead collisions at the Large Hadron Collider make gold at up to about 89,000 nuclei per second. All of Run 2 (2015 to 2018) produced about 29 picograms, and each nucleus survives only a tiny fraction of a second.
• A July 2025 preprint proposes gold from mercury-198 in a fusion power plant blanket. It is an unproven simulation for a plant that does not exist, and its authors expect years of storage before the gold's radioactivity falls to everyday levels.
• Natural gold is entirely gold-197, according to NIST. Most transmutation routes make other, radioactive gold isotopes.
• "Lab-grown gold" does not exist the way lab-grown diamond does. US labels follow the FTC Jewelry Guides (16 CFR Part 23); the UK requires independent assay and a hallmark.

A dated timeline of making gold

Why can't chemistry turn lead into gold?

What makes gold gold

Every atom has a nucleus made of protons and neutrons, surrounded by electrons. The number of protons, called the atomic number, decides which element the atom is. NIST's atomic weights tables list gold at atomic number 79, mercury at 80 and bismuth at 83. CERN's press release notes that a lead nucleus contains 82 protons. Add or remove one proton and you have a different element, full stop.

The neutron count can vary without changing the element. Those variants are isotopes. NIST lists exactly one isotope in natural gold: gold-197, with an isotopic composition of 1, meaning 100 percent. Mercury, by contrast, has seven naturally occurring isotopes in NIST's table, from mercury-196 at about 0.15 percent to mercury-202 at about 29.9 percent. Mercury-198 makes up roughly 10 percent. Bismuth, like gold, has a single natural isotope, bismuth-209.

So "turning lead into gold" means removing three protons from each lead nucleus. Turning mercury into gold means removing one. Turning bismuth into gold means removing four.

Chemistry moves electrons, not protons

A chemical reaction breaks and forms bonds by sharing or transferring electrons. Burning, dissolving, alloying and plating all work at that level, while the nucleus, held together by the strong nuclear force, sits untouched. Nuclear binding energies are roughly a million times larger than chemical bond energies, so no flame, acid or catalyst comes close to rearranging protons. CERN's account of the ALICE result says plainly that "chemical methods are powerless" here. Chemistry can make something that looks like gold or is coated in gold, never new gold atoms.

Density and colour fooled people for centuries

CERN suggests alchemists chased lead partly because it is of a similar density to gold. Density and colour are what people check by hand, and both can be faked, which is why assay and hallmarking exist.

A short history of alchemy and chrysopoeia

The quest is called chrysopoeia, Greek for "gold making". CERN calls it a dream of medieval alchemists; the fusion preprint notes it "has been pursued for millennia". Alchemists in the Hellenistic world, the Islamic world and medieval Europe tried to purify base metals into gold. They failed for the reason above, but left distillation, acids and laboratory habits that chemistry inherited. Some also faked demonstrations with gilding or hidden gold, a pattern modern scams still follow.

The turn came with radioactivity. Ernest Rutherford received the 1908 Nobel Prize in Chemistry "for his investigations into the disintegration of the elements, and the chemistry of radioactive substances." Transmutation was real after all; it simply happens in the nucleus, not in a furnace.

What is nuclear transmutation?

Nuclear transmutation is any process that changes the number of protons in a nucleus, turning one element into another. CERN's release summarises the two broad ways: naturally, by radioactive decay, or in the laboratory, "under a bombardment of neutrons or protons."

In outline:

• Neutron reactions. In an (n,p) reaction a nucleus absorbs a neutron and ejects a proton, so mercury (80) becomes gold (79). In an (n,2n) reaction one neutron goes in and two come out, changing the isotope; the new isotope may then decay into another element.
• Heavy ion collisions. A fast heavy nucleus knocks pieces off a target nucleus, leaving fragments that include gold when the target is bismuth.
• Photon induced reactions. An energetic photon makes a nucleus shed neutrons and protons. At the LHC, a passing lead nucleus's electromagnetic field supplies the photons.

All of these need accelerators, research reactors or a fusion plant, run by licensed facilities under radiation safety rules.

Documented cases of man-made gold

1941: mercury to gold with fast neutrons at Harvard

The earliest well documented paper is "Transmutation of Mercury by Fast Neutrons" by R. Sherr, K. T. Bainbridge and H. H. Anderson of Harvard University, received by Physical Review on 13 August 1941 and published on 1 October 1941 (volume 60, page 473).

According to the abstract, the team irradiated mercury with fast neutrons produced by bombarding lithium with deuterons. They found three radioactive gold isotopes formed by (n,p) reactions: gold-198 with a 65 hour period, gold-199 with a 78 hour period, and a new, then unassigned electron emitter with a 48 minute period. They also found radioactive platinum isotopes, made when a mercury nucleus absorbed a neutron and ejected an alpha particle, and radioactive mercury isotopes.

The gold was a trace inside a mercury target, identified by its radiation rather than weighed, and none of it was stable gold-197. With half-lives under four days, it decayed away within weeks.

1980: bismuth to gold at Berkeley

The experiment most often described as "Seaborg turned bismuth into gold" is published as "Energy dependence of bismuth-209 fragmentation in relativistic nuclear collisions" by K. Aleklett, D. J. Morrissey, W. Loveland, P. L. McGaughey and G. T. Seaborg of the Nuclear Science Division, Lawrence Berkeley Laboratory. Physical Review C received it on 5 August 1980 and published it on 1 March 1981 (volume 23, page 1044). Glenn Seaborg had shared the 1951 Nobel Prize in Chemistry for discoveries in the chemistry of transuranium elements, which gave the story a famous name.

The abstract reports bismuth-209 struck by carbon-12 ions at 4.8 and 25.2 GeV and neon-20 ions at 8.0 GeV, with the gold isotopes identified by germanium detector gamma spectroscopy. In other words, the gold was found by the gamma rays it emitted as it decayed: radioactive traces, not an ingot. The scientific point was how heavy nuclei fragment at relativistic energies. Press coverage at the time quoted large cost estimates that we could not trace to a primary source, so we do not repeat them.

2025: lead to gold at the Large Hadron Collider

On 8 May 2025, CERN announced that the ALICE collaboration had measured the transmutation of lead into gold at the Large Hadron Collider. The underlying paper, "Proton emission in ultraperipheral Pb-Pb collisions at 5.02 TeV", appeared in Physical Review C volume 111, article 054906, dated 7 May 2025.

What happens, in CERN's words and the paper's abstract:

1. The LHC accelerates lead-208 nuclei to 99.999993 percent of the speed of light.
2. Most interactions are near misses rather than head-on collisions. Each lead nucleus carries 82 protons, and at that speed its electromagnetic field is squashed into a thin pancake that acts like a short, intense pulse of photons.
3. When a photon from one nucleus hits the other, it can excite the nucleus so that it ejects a few neutrons and protons. This is called electromagnetic dissociation.
4. Removing three protons (plus at least one neutron) turns lead into gold. The illustration in CERN's release shows two neutrons and three protons leaving, which leaves gold-203.

ALICE used its zero degree calorimeters to count events in which zero, one, two or three protons were emitted together with at least one neutron. Those correspond to production of lead, thallium, mercury and gold isotopes respectively. The paper reports that a theoretical model (RELDIS) describes the zero proton and three proton cross sections within uncertainties, while underestimating the one and two proton channels by 17 to 25 percent.

The headline figures from CERN:

• Gold is produced at a maximum rate of about 89,000 nuclei per second from lead-lead collisions at the ALICE collision point.
• During Run 2 (2015 to 2018), about 86 billion gold nuclei were created at the four major experiments. That corresponds to 29 picograms, or 2.9 × 10^-11 grams.
• Run 3 has produced almost double the Run 2 amount, still "trillions of times less" than a piece of jewellery needs.
• The gold nuclei leave the collision with very high energy, hit the beam pipe or collimators downstream, and immediately fragment into protons, neutrons and other particles. The gold exists for just a tiny fraction of a second.

The payoff is physics: these processes help predict beam losses, a major limit on collider performance. As CERN put it, the alchemists' hopes of riches "have once again been dashed."

The methods side by side

Could fusion power plants make gold from mercury?

This is the only proposal that claims gold at an economically interesting scale, so it deserves a careful, labelled look.

What is claimed. A preprint titled "Scalable Chrysopoeia via (n, 2n) Reactions Driven by Deuterium-Tritium Fusion Neutrons" by Adam Rutkowski, Jake Harter and Jason Parisi of Marathon Fusion, San Francisco, was posted to arXiv on 17 July 2025, with a third version on 3 August 2025. It proposes placing mercury enriched in mercury-198 in a neutron multiplier layer of a fusion power plant blanket. Fast 14 MeV neutrons from a deuterium-tritium plasma would knock a neutron out of mercury-198, making mercury-197, which then decays to stable gold-197. Using neutronics simulations, the authors report a rate of about 2 tonnes of gold per gigawatt of thermal power per year, and argue this could double a fusion plant's revenue without hurting electricity output or tritium breeding.

Why it is unproven. It is an arXiv preprint with no peer reviewed version we could find. The result comes from simulation, not weighable gold. It assumes a deuterium-tritium fusion power plant with a working breeding blanket, and no such commercial plant operates today. It also needs large amounts of isotope enriched mercury inside the blanket.

The radioactivity problem, in the authors' own numbers. Side reactions on mercury-196 lead to gold-195, which the preprint gives a half-life of 186.10 days. For the modelled isotope mix the authors estimate:

• about 6.8 years of storage before the activity falls below the Nuclear Regulatory Commission's Class A low level waste limit the paper cites;
• about 13.7 years before the gold falls below the activity threshold at which it would need labelling as radioactive for transport, by the authors' reading of the rules;
• about 17.7 years before a troy ounce would be less radioactive than a banana, a deliberately conservative benchmark the authors chose.

The authors argue such storage is acceptable because much gold is held as a store of value. Until a fusion plant runs and someone produces, cools and assays the metal, this stays unproven.

How much would artificial gold cost per gram?

Start with the market. Bernama reported that physical gold was fixed at US$4,089.25 per troy ounce at the LBMA afternoon price on 14 October 2026. NIST's conversion tables give one troy ounce as 31.10348 grams. That puts gold at roughly US$131 per gram in mid October 2026. Gold prices move every day, so treat this as a snapshot.

Now the LHC, the best measured modern case. The Python below uses only numbers from the sources: the CERN Run 2 mass, the peak rate at ALICE, the NIST troy ounce and the reported price. It also uses the atomic mass unit and assumes gold-203, the isotope shown in CERN's illustration.

# Value of the gold the LHC made, using published figures
TROY_OZ_G = 31.10348          # NIST SP 811, Appendix B
PRICE_PER_OZ = 4089.25        # LBMA PM fix, 14 Oct 2026, as reported by Bernama
price_per_g = PRICE_PER_OZ / TROY_OZ_G
print(f"Gold price: ${price_per_g:,.2f} per gram")

run2_mass_g = 29e-12          # CERN: about 29 picograms in Run 2 (2015 to 2018)
print(f"Run 2 gold value: ${run2_mass_g * price_per_g:.2e}")

# Peak production at the ALICE point: about 89,000 nuclei per second
AMU_G = 1.66054e-24           # grams per atomic mass unit
nucleus_g = 203 * AMU_G       # gold-203, per CERN's illustration
grams_per_second = 89_000 * nucleus_g
seconds_for_one_gram = 1 / grams_per_second
years = seconds_for_one_gram / (365.25 * 24 * 3600)
print(f"Years at peak rate to make 1 gram: {years:,.2e}")

Running it prints a gold price of about US$131.47 per gram, a Run 2 value of about US$3.8 × 10^-9 (well under a millionth of a cent), and roughly 1.06 × 10^9 years of nonstop running at the peak rate to accumulate one gram. That billion year figure is generous in two ways: the LHC only runs lead beams for a few weeks in most years, and the gold does not accumulate at all, because each nucleus fragments almost immediately.

The 1941 and 1980 experiments tell the same story: costly beam time for a trace of radioactive gold. No published case made gold for less than mined gold costs. Only the unproven fusion preprint claims profitability, and it assumes the plant already pays for itself through electricity.

Why is man-made gold usually radioactive?

Stable nuclei need a particular balance of protons and neutrons, and for gold only one combination occurs in nature: 79 protons and 118 neutrons, gold-197. Transmutation rarely lands there. The 1941 neutrons made gold-198 and gold-199, the 1980 bismuth fragments were found by their gamma rays, and the LHC's lead collisions give isotopes such as gold-203. The fusion proposal aims at gold-197 via mercury-197 decay, but side reactions make radioactive gold-195, hence the years of cooling. The target and surrounding material also become radioactive, which is why transmutation belongs only in licensed facilities.

What is "lab-grown gold"?

Lab-grown diamonds exist because diamond is a crystal structure of carbon that can be grown from carbon. Gold is an element, so you cannot grow it from anything that is not already gold, except by the nuclear routes above. A listing for "lab gold", "synthetic gold" or "alchemy gold" honestly means one of these:

• Recycled or refined gold. Real gold recovered from scrap or electronics. Nothing artificial about the atoms.
• A gold alloy. Real gold mixed with other metals; 14 karat is 14 parts gold to 10 parts other metal, as the FTC explains.
• Gold plated, gold filled or vermeil. A base metal or silver core with a gold layer.
• A gold coloured material with no gold. Brass or golden coatings on steel or titanium.

Each can be a fine product at a fair price. The problem is selling one as solid gold.

Plated, filled, vermeil: what the US labels mean

In the United States, the Federal Trade Commission's Guides for the Jewelry, Precious Metals, and Pewter Industries (16 CFR Part 23) set out what is unfair or deceptive in describing gold. The FTC says it is unfair or deceptive to misrepresent the presence of gold, its karat fineness, or the thickness, weight ratio or manner of application of any gold plating or coating. The key rules, taken from the current eCFR text of section 23.3 and section 23.4 and the FTC's consumer guide:

A footnote in the guides adds a useful conversion: 1 micron of 12 karat gold is equivalent to half a micron of 24 karat gold. Another footnote cites the National Stamping Act tolerance for gold: three parts per thousand for articles without solder and seven parts per thousand where the tested part contains solder.

The FTC's consumer page adds practical advice. Look for a karat mark, and for the name or US registered trademark of the company that stands behind it. In the FTC's words, if you do not see a trademark along with the karat mark, "don't buy it." It also warns that "gold flashed" and "gold washed" describe a very thin electroplating that wears away faster than other plating.

One nuance worth knowing: older summaries often say US jewellery under 10 karat cannot be called gold. The current text of section 23.3 includes "9 Karat Gold" among its examples of acceptable markings, provided the karat designation is correct and conspicuous. The 10 karat floor in the current text applies to the gold layer in plated, filled, overlay and vermeil products.

Karats, fineness and hallmarks

Karats and parts per thousand

A karat (spelled carat in the UK for gold) is a 24th part. Fineness is the same idea in parts per thousand, which is how hallmarks and many stamps express it.

The 585 stamp for 14 karat sits slightly above the exact 583.3 so that a piece marked 585 is guaranteed to meet the standard. The Goldsmiths' Company Assay Office lists 375, 585, 750, 916 and 999 as the gold finenesses recognised under the international Hallmarking Convention, and its price calculator also offers 990.

UK hallmarking: independent testing by law

The UK system is the oldest continuous consumer protection scheme for precious metals. The Goldsmiths' Company Assay Office in London says it has tested and hallmarked precious metals since 1300. Under the Hallmarking Act 1973, it is an offence in the course of trade to describe an unhallmarked article as wholly or partly precious metal, or to supply one so described. The Assay Office says Trading Standards can seize non compliant items, with fines of up to £5,000 per article. Weight exemptions apply; for gold, the Assay Office says items over 1 gram need a UK recognised hallmark.

A full traditional UK hallmark has five parts, three of them compulsory:

• Sponsor's mark (compulsory): the initials of the person or company that submitted the item and takes responsibility for it, inside a shield.
• Traditional fineness symbol (optional): a symbol for the metal type.
• Millesimal fineness mark (compulsory since 1999): the precious metal content in parts per thousand, inside a shield whose shape indicates the metal.
• Assay office mark (compulsory): London's leopard's head, or the marks of Birmingham, Sheffield or Edinburgh.
• Date letter (optional): changes every 1 January.

The Assay Office stresses the reason the system exists: precious metal content cannot be detected by sight or touch. It marks a piece to the lowest standard found, so the whole article is guaranteed to be no less fine than the mark says.

US marks: manufacturer responsibility, not independent assay

In the US, a karat stamp is the manufacturer's or seller's claim, backed by the FTC Guides and the National Stamping Act, which is why the FTC wants a trademark beside it: it tells you who stands behind the claim. Other countries run their own assay systems, so check the rules behind any foreign mark.

Step by step: check what a gold item really is

A non destructive buyer's checklist using your eyes, a loupe and documents. No acids or chemical tests; leave those to jewellers and assay offices.

1. Find the marks. Look on clasps, inside ring shanks and on tags. Note the karat or fineness (for example 14K or 585) and any trademark or sponsor's mark next to it.
2. Check for qualifiers. Look for GP, GEP, HGE, RGP, GF, a fraction such as 1/20, or the word vermeil. Any of these means the item is plated or filled, not solid karat gold throughout.
3. Match the mark to the claim. A listing that says "solid 18K gold" should carry an 18K or 750 mark. A listing that says "gold" with no karat should, under the FTC Guides, mean 24 karat throughout. Anything else is a red flag.
4. For UK items over the weight threshold, look for the compulsory marks. Sponsor's mark, millesimal fineness and an assay office mark. If they are missing on an item described as gold, it should not be sold as gold in the UK.
5. Ask for documents. Request the seller's name, the trademark owner, the composition in writing, and the refund and return policy, as the FTC suggests.
6. Do the arithmetic on value. Gold content times weight times the day's price sets a floor for the metal value. Use the small script below; if the asking price is far below metal value, something is wrong.
7. When it matters, get it tested. For valuable items, use a jeweller with non destructive testing equipment, or an assay office service. The London Assay Office, for example, offers laboratory services and hallmark authentication.

# Rough metal value of a gold item (ignores making charges, gems and dealer margins)
def gold_value(weight_g: float, karat: float, price_per_troy_oz: float) -> float:
    fineness = karat / 24
    price_per_g = price_per_troy_oz / 31.10348   # NIST troy ounce
    return weight_g * fineness * price_per_g

# Example: a 5 gram 14K chain at the 14 Oct 2026 LBMA PM fix
print(round(gold_value(5.0, 14, 4089.25), 2))   # about 383.46

A gold plated chain of the same weight carries only a microns thin layer of gold, so its metal value is a tiny fraction of that number.

Common mistakes: misconceptions and scams

"Chemists just haven't found the right reaction." There is no right reaction. Chemistry changes how atoms bond; it cannot change how many protons a nucleus holds. This is settled physics, not a gap waiting for a clever recipe.

"CERN can make gold, so gold will get cheap." CERN's own release says Run 3's output is still trillions of times less than a piece of jewellery needs, and the nuclei fragment within a tiny fraction of a second. The LHC is a physics instrument; the gold is a byproduct measured to understand beam losses.

"Fusion gold is coming next year." The mercury-198 proposal is a simulation in a preprint, for a fusion plant that does not exist. Its own authors estimate years of cooling before the gold is ordinary. Anyone selling shares, tokens or presales on "fusion gold" today is selling a promise, not a product.

"Man-made gold would be identical to mined gold, and lab-grown gold is like lab-grown diamond." Most documented routes made radioactive isotopes, not gold-197. And "lab gold" listings are plated, alloyed, recycled or gold coloured items with marketing on top.

"Gold plated means it has real gold, so it's worth something." It does have some gold, but an electroplate can be as thin as 0.175 microns under the FTC minimum, and "gold flashed" or "gold washed" can be thinner still. The metal value is usually negligible.

"The density test proves it's gold." CERN points out that lead and gold have similar densities, which is part of why alchemists chased lead. Dense base metal cores under a gold layer can pass a casual weight check. Trust marks, documents and proper testing.

"Precious metals deals are outside regulation." The CFTC lists this claim as a fraud warning sign, along with guaranteed profits, invented scarcity, financed purchases where you pay 15 to 25 percent and owe the rest, and offers to store metal in a "bank" or overseas facility that may hold nothing. Check registration with the CFTC or the National Futures Association first.

"An old alchemy secret, rediscovered." Any claim to have turned base metal into gold chemically is false.

FAQ

Can you turn lead into gold?

Not with chemistry. With nuclear physics, yes, in trace amounts: CERN's ALICE experiment reported in May 2025 that near-miss lead collisions at the LHC produce gold nuclei, about 29 picograms in all of Run 2, which fragment almost immediately.

Has anyone actually made gold?

Yes: Harvard from mercury in 1941, a Berkeley team including Glenn Seaborg from bismuth (published 1981), and ALICE from lead in 2025. All made tiny, mostly radioactive amounts.

How much does it cost to make gold artificially?

Far more than gold is worth. At the LHC's peak rate at ALICE, making one gram would take on the order of a billion years of nonstop running, while mid October 2026 market gold sold for roughly US$131 per gram.

Is artificially made gold radioactive?

Usually. The 1941 experiment made gold-198 and gold-199, the 1980 experiment measured gold isotopes by their gamma rays, and the 2025 fusion proposal's authors estimate years of storage before their product's radioactivity falls to everyday levels.

Can nuclear fusion turn mercury into gold?

A July 2025 preprint from Marathon Fusion argues it could, at about 2 tonnes per gigawatt thermal per year, using simulations. It is not peer reviewed and needs a fusion power plant that does not exist yet, so it is unproven.

Is lab-grown gold real gold?

There is no lab-grown gold in the sense of lab-grown diamond. Products sold that way are recycled gold, gold alloys, plated or filled items, or gold coloured materials with no gold.

What is the difference between gold plated, gold filled and vermeil?

Under the FTC Guides, gold electroplate needs at least 10 karat gold at least 0.175 microns thick; gold filled needs a mechanically bonded 10 karat or better layer making up at least 1/20 of the metal weight; vermeil is sterling silver with at least 2.5 microns of 10 karat or better gold.

What does 585 or 750 stamped on gold mean?

It is the fineness in parts per thousand: 585 is 14 karat and 750 is 18 karat. In the UK, a millesimal fineness mark has been a compulsory part of the hallmark since 1999.

Fix it without waiting for upstream

The alchemists wanted to change what a metal is at its core and only managed to change its surface; with software, the opposite trap is common, where a component looks current on the outside while a known vulnerability sits deep in its dependency tree. Safeguard Gold is a different kind of transformation: it takes the open source components you already depend on and gives you hardened, verified builds, with signatures and evidence playing the role a hallmark plays for metal.

Whatever the vulnerability, most of the risk sits in two gaps. The first is the time between a fix being published and that fix actually running in your production builds. The second, and worse, is the time before any fix exists at all, while a maintainer works out a patch and a release schedule. Waiting through both gaps is a habit, not a requirement. Safeguard gives teams three ways to close them.

1. Autonomous remediation: pull requests, not tickets

When a fixed version exists, Safeguard's Griffin models open a tested, policy gated pull request that moves the affected dependency to the safe version. You can run it on one repository or in bulk across every project that carries the vulnerable version. Remediation covers eight categories, including dependencies, licenses, security posture and findings. Reachability analysis ranks the vulnerable code paths your application actually calls first, so the first pull requests you merge are the ones that matter most. Autonomous remediation ships switched off, with every category set to Manual, so your team decides how much runs on its own and what still needs a human approval.

2. Gold Open Source: the safe version in seconds, free

gold.safeguard.sh is free and needs no login. Search any CVE or package to see affected and fixed versions, CVSS, EPSS, CISA KEV status and a recommended safe version, or use the "Am I affected?" checker. The same data is available through a free read API, README status badges and RSS threat feeds. The safeguard-sh/gold-scan-action GitHub Action can fail a build on a known exploited vulnerability, on any known vulnerability, or never, so a version you have just removed cannot quietly come back in next week's merge.

3. Gold Registry: start from components that are already fixed

registry.safeguard.sh provides hardened, drop in replacements for upstream packages and container images. Every Gold artifact is held continuously to zero exploitable critical or high CVEs, zero malware and a completed license check, with SLSA L3 or higher provenance, a Sigstore signature and a public evidence page you can hand to an auditor. An artifact loses its Gold badge if it regresses. Teams can point their own package directory at the registry so hardened equivalents resolve without changing how developers install anything.

When upstream has no fix yet, or will not ship one, Safeguard first works to get the fix accepted upstream. If the issue is severe or upstream will not move, Safeguard builds a fork, fixes the vulnerability there and publishes a hardened Gold build, so you are not stuck on a maintainer's release schedule. You can also request a hardened, compatibility tested build of a specific package. This fork and patch flow is in early access today, so ask which packages it covers before you plan around it.

One honest limit: when the vulnerable code lives in a vendor appliance, a hosted service or an operating system runtime you do not build yourself, the fix still has to come from that vendor or distribution. The steps above apply to the open source dependencies and container images you build and ship, which is where most teams carry the largest and least tracked share of their exposure.

If the hallmark analogy appeals, our explainers on software provenance, signing with Sigstore and cosign, reproducible builds and how the Gold build pipeline works cover the software version of "independently tested, and you can see who stands behind it." To try it on your own machine, download Safeguard from https://safeguard.sh/download and see the CLI documentation at https://docs.safeguard.sh/docs/cli .

See how Safeguard would help

To be plain about the fit: Safeguard has nothing to do with physics or assaying. These demos show the software side of the analogy.

The first demo is a real Safeguard CLI session. Asked to fix a lodash CVE, it reads the finding and advisory, asks before each command, edits package.json to lodash 4.17.21, runs npm install and npm test with approval, prints a summary and pushes nothing. Think of it as replacing a plated part with a solid one, with you signing off at every step.

Click through it at your own pace, or open it full screen.

The second demo sets up Safeguard's local runner on your own machine, approves it with its device code, and then scans from the terminal. It is the software equivalent of reading the marks before you buy: you see which components and versions are actually in a project.

Click through it at your own pace, or open it full screen.

The third demo connects Claude to the Safeguard MCP server with OAuth and lets you choose which tools Claude can use. That lets you ask questions such as which projects carry a given package version, while you control exactly what the assistant can touch.

Click through it at your own pace, or open it full screen.

Quick check

Sources

• CERN: ALICE detects the conversion of lead into gold at the LHC, 8 May 2025
• ALICE Collaboration, Proton emission in ultraperipheral Pb-Pb collisions at 5.02 TeV, Physical Review C 111, 054906 (2025)
• Sherr, Bainbridge and Anderson, Transmutation of Mercury by Fast Neutrons, Physical Review 60, 473 (1941)
• Aleklett, Morrissey, Loveland, McGaughey and Seaborg, Energy dependence of bismuth-209 fragmentation in relativistic nuclear collisions, Physical Review C 23, 1044 (1981)
• Rutkowski, Harter and Parisi, Scalable Chrysopoeia via (n, 2n) Reactions Driven by Deuterium-Tritium Fusion Neutrons, arXiv:2507.13461 (preprint, not peer reviewed)
• Nobel Prize in Chemistry 1908: Ernest Rutherford, nobelprize.org
• NIST: Atomic weights and isotopic compositions for gold
• NIST: Atomic weights and isotopic compositions for mercury
• NIST: Atomic weights and isotopic compositions for bismuth
• NIST Special Publication 811, Appendix B.9: conversion factors (troy ounce)
• eCFR: 16 CFR Part 23, Guides for the Jewelry, Precious Metals, and Pewter Industries
• FTC consumer advice: Buying platinum, gold, and silver jewelry
• CFTC: Precious metals fraud advisory
• Goldsmiths' Company Assay Office: What is a hallmark?
• Goldsmiths' Company Assay Office: Why hallmark?
• Goldsmiths' Company Assay Office: Hallmarking
• Bernama: Gold futures end higher, tracking COMEX gains, 15 October 2026 (reports the LBMA PM fix for 14 October 2026)

Last reviewed: 23 October 2026.